Effective Feature Preprocessing for Time Series Forecasting
DEFF Research Database (Denmark)
Zhao, Junhua; Dong, Zhaoyang; Xu, Zhao
2006-01-01
Time series forecasting is an important area in data mining research. Feature preprocessing techniques have significant influence on forecasting accuracy, therefore are essential in a forecasting model. Although several feature preprocessing techniques have been applied in time series forecasting...... performance in time series forecasting. It is demonstrated in our experiment that, effective feature preprocessing can significantly enhance forecasting accuracy. This research can be a useful guidance for researchers on effectively selecting feature preprocessing techniques and integrating them with time...... series forecasting models....
Adaptive time-variant models for fuzzy-time-series forecasting.
Wong, Wai-Keung; Bai, Enjian; Chu, Alice Wai-Ching
2010-12-01
A fuzzy time series has been applied to the prediction of enrollment, temperature, stock indices, and other domains. Related studies mainly focus on three factors, namely, the partition of discourse, the content of forecasting rules, and the methods of defuzzification, all of which greatly influence the prediction accuracy of forecasting models. These studies use fixed analysis window sizes for forecasting. In this paper, an adaptive time-variant fuzzy-time-series forecasting model (ATVF) is proposed to improve forecasting accuracy. The proposed model automatically adapts the analysis window size of fuzzy time series based on the prediction accuracy in the training phase and uses heuristic rules to generate forecasting values in the testing phase. The performance of the ATVF model is tested using both simulated and actual time series including the enrollments at the University of Alabama, Tuscaloosa, and the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX). The experiment results show that the proposed ATVF model achieves a significant improvement in forecasting accuracy as compared to other fuzzy-time-series forecasting models.
Neural Network Models for Time Series Forecasts
Tim Hill; Marcus O'Connor; William Remus
1996-01-01
Neural networks have been advocated as an alternative to traditional statistical forecasting methods. In the present experiment, time series forecasts produced by neural networks are compared with forecasts from six statistical time series methods generated in a major forecasting competition (Makridakis et al. [Makridakis, S., A. Anderson, R. Carbone, R. Fildes, M. Hibon, R. Lewandowski, J. Newton, E. Parzen, R. Winkler. 1982. The accuracy of extrapolation (time series) methods: Results of a ...
Introduction to time series analysis and forecasting
Montgomery, Douglas C; Kulahci, Murat
2008-01-01
An accessible introduction to the most current thinking in and practicality of forecasting techniques in the context of time-oriented data. Analyzing time-oriented data and forecasting are among the most important problems that analysts face across many fields, ranging from finance and economics to production operations and the natural sciences. As a result, there is a widespread need for large groups of people in a variety of fields to understand the basic concepts of time series analysis and forecasting. Introduction to Time Series Analysis and Forecasting presents the time series analysis branch of applied statistics as the underlying methodology for developing practical forecasts, and it also bridges the gap between theory and practice by equipping readers with the tools needed to analyze time-oriented data and construct useful, short- to medium-term, statistically based forecasts.
Forecasting Enrollments with Fuzzy Time Series.
Song, Qiang; Chissom, Brad S.
The concept of fuzzy time series is introduced and used to forecast the enrollment of a university. Fuzzy time series, an aspect of fuzzy set theory, forecasts enrollment using a first-order time-invariant model. To evaluate the model, the conventional linear regression technique is applied and the predicted values obtained are compared to the…
25 years of time series forecasting
de Gooijer, J.G.; Hyndman, R.J.
2006-01-01
We review the past 25 years of research into time series forecasting. In this silver jubilee issue, we naturally highlight results published in journals managed by the International Institute of Forecasters (Journal of Forecasting 1982-1985 and International Journal of Forecasting 1985-2005). During
Neural network versus classical time series forecasting models
Nor, Maria Elena; Safuan, Hamizah Mohd; Shab, Noorzehan Fazahiyah Md; Asrul, Mohd; Abdullah, Affendi; Mohamad, Nurul Asmaa Izzati; Lee, Muhammad Hisyam
2017-05-01
Artificial neural network (ANN) has advantage in time series forecasting as it has potential to solve complex forecasting problems. This is because ANN is data driven approach which able to be trained to map past values of a time series. In this study the forecast performance between neural network and classical time series forecasting method namely seasonal autoregressive integrated moving average models was being compared by utilizing gold price data. Moreover, the effect of different data preprocessing on the forecast performance of neural network being examined. The forecast accuracy was evaluated using mean absolute deviation, root mean square error and mean absolute percentage error. It was found that ANN produced the most accurate forecast when Box-Cox transformation was used as data preprocessing.
Time Series Forecasting with Missing Values
Directory of Open Access Journals (Sweden)
Shin-Fu Wu
2015-11-01
Full Text Available Time series prediction has become more popular in various kinds of applications such as weather prediction, control engineering, financial analysis, industrial monitoring, etc. To deal with real-world problems, we are often faced with missing values in the data due to sensor malfunctions or human errors. Traditionally, the missing values are simply omitted or replaced by means of imputation methods. However, omitting those missing values may cause temporal discontinuity. Imputation methods, on the other hand, may alter the original time series. In this study, we propose a novel forecasting method based on least squares support vector machine (LSSVM. We employ the input patterns with the temporal information which is defined as local time index (LTI. Time series data as well as local time indexes are fed to LSSVM for doing forecasting without imputation. We compare the forecasting performance of our method with other imputation methods. Experimental results show that the proposed method is promising and is worth further investigations.
Robust Forecasting of Non-Stationary Time Series
Croux, C.; Fried, R.; Gijbels, I.; Mahieu, K.
2010-01-01
This paper proposes a robust forecasting method for non-stationary time series. The time series is modelled using non-parametric heteroscedastic regression, and fitted by a localized MM-estimator, combining high robustness and large efficiency. The proposed method is shown to produce reliable forecasts in the presence of outliers, non-linearity, and heteroscedasticity. In the absence of outliers, the forecasts are only slightly less precise than those based on a localized Least Squares estima...
Forecasting Cryptocurrencies Financial Time Series
DEFF Research Database (Denmark)
Catania, Leopoldo; Grassi, Stefano; Ravazzolo, Francesco
2018-01-01
This paper studies the predictability of cryptocurrencies time series. We compare several alternative univariate and multivariate models in point and density forecasting of four of the most capitalized series: Bitcoin, Litecoin, Ripple and Ethereum. We apply a set of crypto–predictors and rely...
Forecasting with nonlinear time series models
DEFF Research Database (Denmark)
Kock, Anders Bredahl; Teräsvirta, Timo
In this paper, nonlinear models are restricted to mean nonlinear parametric models. Several such models popular in time series econo- metrics are presented and some of their properties discussed. This in- cludes two models based on universal approximators: the Kolmogorov- Gabor polynomial model...... applied to economic fore- casting problems, is briefly highlighted. A number of large published studies comparing macroeconomic forecasts obtained using different time series models are discussed, and the paper also contains a small simulation study comparing recursive and direct forecasts in a partic...... and two versions of a simple artificial neural network model. Techniques for generating multi-period forecasts from nonlinear models recursively are considered, and the direct (non-recursive) method for this purpose is mentioned as well. Forecasting with com- plex dynamic systems, albeit less frequently...
Layered Ensemble Architecture for Time Series Forecasting.
Rahman, Md Mustafizur; Islam, Md Monirul; Murase, Kazuyuki; Yao, Xin
2016-01-01
Time series forecasting (TSF) has been widely used in many application areas such as science, engineering, and finance. The phenomena generating time series are usually unknown and information available for forecasting is only limited to the past values of the series. It is, therefore, necessary to use an appropriate number of past values, termed lag, for forecasting. This paper proposes a layered ensemble architecture (LEA) for TSF problems. Our LEA consists of two layers, each of which uses an ensemble of multilayer perceptron (MLP) networks. While the first ensemble layer tries to find an appropriate lag, the second ensemble layer employs the obtained lag for forecasting. Unlike most previous work on TSF, the proposed architecture considers both accuracy and diversity of the individual networks in constructing an ensemble. LEA trains different networks in the ensemble by using different training sets with an aim of maintaining diversity among the networks. However, it uses the appropriate lag and combines the best trained networks to construct the ensemble. This indicates LEAs emphasis on accuracy of the networks. The proposed architecture has been tested extensively on time series data of neural network (NN)3 and NN5 competitions. It has also been tested on several standard benchmark time series data. In terms of forecasting accuracy, our experimental results have revealed clearly that LEA is better than other ensemble and nonensemble methods.
Variable Selection in Time Series Forecasting Using Random Forests
Directory of Open Access Journals (Sweden)
Hristos Tyralis
2017-10-01
Full Text Available Time series forecasting using machine learning algorithms has gained popularity recently. Random forest is a machine learning algorithm implemented in time series forecasting; however, most of its forecasting properties have remained unexplored. Here we focus on assessing the performance of random forests in one-step forecasting using two large datasets of short time series with the aim to suggest an optimal set of predictor variables. Furthermore, we compare its performance to benchmarking methods. The first dataset is composed by 16,000 simulated time series from a variety of Autoregressive Fractionally Integrated Moving Average (ARFIMA models. The second dataset consists of 135 mean annual temperature time series. The highest predictive performance of RF is observed when using a low number of recent lagged predictor variables. This outcome could be useful in relevant future applications, with the prospect to achieve higher predictive accuracy.
forecasting with nonlinear time series model: a monte-carlo
African Journals Online (AJOL)
PUBLICATIONS1
erated recursively up to any step greater than one. For nonlinear time series model, point forecast for step one can be done easily like in the linear case but forecast for a step greater than or equal to ..... London. Franses, P. H. (1998). Time series models for business and Economic forecasting, Cam- bridge University press.
Forecasting Cryptocurrencies Financial Time Series
Catania, Leopoldo; Grassi, Stefano; Ravazzolo, Francesco
2018-01-01
This paper studies the predictability of cryptocurrencies time series. We compare several alternative univariate and multivariate models in point and density forecasting of four of the most capitalized series: Bitcoin, Litecoin, Ripple and Ethereum. We apply a set of crypto–predictors and rely on Dynamic Model Averaging to combine a large set of univariate Dynamic Linear Models and several multivariate Vector Autoregressive models with different forms of time variation. We find statistical si...
Introduction to time series and forecasting
Brockwell, Peter J
2016-01-01
This book is aimed at the reader who wishes to gain a working knowledge of time series and forecasting methods as applied to economics, engineering and the natural and social sciences. It assumes knowledge only of basic calculus, matrix algebra and elementary statistics. This third edition contains detailed instructions for the use of the professional version of the Windows-based computer package ITSM2000, now available as a free download from the Springer Extras website. The logic and tools of time series model-building are developed in detail. Numerous exercises are included and the software can be used to analyze and forecast data sets of the user's own choosing. The book can also be used in conjunction with other time series packages such as those included in R. The programs in ITSM2000 however are menu-driven and can be used with minimal investment of time in the computational details. The core of the book covers stationary processes, ARMA and ARIMA processes, multivariate time series and state-space mod...
Introduction to time series analysis and forecasting
Montgomery, Douglas C; Kulahci, Murat
2015-01-01
Praise for the First Edition ""…[t]he book is great for readers who need to apply the methods and models presented but have little background in mathematics and statistics."" -MAA Reviews Thoroughly updated throughout, Introduction to Time Series Analysis and Forecasting, Second Edition presents the underlying theories of time series analysis that are needed to analyze time-oriented data and construct real-world short- to medium-term statistical forecasts. Authored by highly-experienced academics and professionals in engineering statistics, the Second Edition features discussions on both
Time Series Analysis and Forecasting by Example
Bisgaard, Soren
2011-01-01
An intuition-based approach enables you to master time series analysis with ease Time Series Analysis and Forecasting by Example provides the fundamental techniques in time series analysis using various examples. By introducing necessary theory through examples that showcase the discussed topics, the authors successfully help readers develop an intuitive understanding of seemingly complicated time series models and their implications. The book presents methodologies for time series analysis in a simplified, example-based approach. Using graphics, the authors discuss each presented example in
Robust Forecasting of Non-Stationary Time Series
Croux, C.; Fried, R.; Gijbels, I.; Mahieu, K.
2010-01-01
This paper proposes a robust forecasting method for non-stationary time series. The time series is modelled using non-parametric heteroscedastic regression, and fitted by a localized MM-estimator, combining high robustness and large efficiency. The proposed method is shown to produce reliable
Forecasting with periodic autoregressive time series models
Ph.H.B.F. Franses (Philip Hans); R. Paap (Richard)
1999-01-01
textabstractThis paper is concerned with forecasting univariate seasonal time series data using periodic autoregressive models. We show how one should account for unit roots and deterministic terms when generating out-of-sample forecasts. We illustrate the models for various quarterly UK consumption
Fuzzy time-series based on Fibonacci sequence for stock price forecasting
Chen, Tai-Liang; Cheng, Ching-Hsue; Jong Teoh, Hia
2007-07-01
Time-series models have been utilized to make reasonably accurate predictions in the areas of stock price movements, academic enrollments, weather, etc. For promoting the forecasting performance of fuzzy time-series models, this paper proposes a new model, which incorporates the concept of the Fibonacci sequence, the framework of Song and Chissom's model and the weighted method of Yu's model. This paper employs a 5-year period TSMC (Taiwan Semiconductor Manufacturing Company) stock price data and a 13-year period of TAIEX (Taiwan Stock Exchange Capitalization Weighted Stock Index) stock index data as experimental datasets. By comparing our forecasting performances with Chen's (Forecasting enrollments based on fuzzy time-series. Fuzzy Sets Syst. 81 (1996) 311-319), Yu's (Weighted fuzzy time-series models for TAIEX forecasting. Physica A 349 (2004) 609-624) and Huarng's (The application of neural networks to forecast fuzzy time series. Physica A 336 (2006) 481-491) models, we conclude that the proposed model surpasses in accuracy these conventional fuzzy time-series models.
Automated time series forecasting for biosurveillance.
Burkom, Howard S; Murphy, Sean Patrick; Shmueli, Galit
2007-09-30
For robust detection performance, traditional control chart monitoring for biosurveillance is based on input data free of trends, day-of-week effects, and other systematic behaviour. Time series forecasting methods may be used to remove this behaviour by subtracting forecasts from observations to form residuals for algorithmic input. We describe three forecast methods and compare their predictive accuracy on each of 16 authentic syndromic data streams. The methods are (1) a non-adaptive regression model using a long historical baseline, (2) an adaptive regression model with a shorter, sliding baseline, and (3) the Holt-Winters method for generalized exponential smoothing. Criteria for comparing the forecasts were the root-mean-square error, the median absolute per cent error (MedAPE), and the median absolute deviation. The median-based criteria showed best overall performance for the Holt-Winters method. The MedAPE measures over the 16 test series averaged 16.5, 11.6, and 9.7 for the non-adaptive regression, adaptive regression, and Holt-Winters methods, respectively. The non-adaptive regression forecasts were degraded by changes in the data behaviour in the fixed baseline period used to compute model coefficients. The mean-based criterion was less conclusive because of the effects of poor forecasts on a small number of calendar holidays. The Holt-Winters method was also most effective at removing serial autocorrelation, with most 1-day-lag autocorrelation coefficients below 0.15. The forecast methods were compared without tuning them to the behaviour of individual series. We achieved improved predictions with such tuning of the Holt-Winters method, but practical use of such improvements for routine surveillance will require reliable data classification methods.
On robust forecasting of autoregressive time series under censoring
Kharin, Y.; Badziahin, I.
2009-01-01
Problems of robust statistical forecasting are considered for autoregressive time series observed under distortions generated by interval censoring. Three types of robust forecasting statistics are developed; meansquare risk is evaluated for the developed forecasting statistics. Numerical results are given.
Forecasting daily meteorological time series using ARIMA and regression models
Murat, Małgorzata; Malinowska, Iwona; Gos, Magdalena; Krzyszczak, Jaromir
2018-04-01
The daily air temperature and precipitation time series recorded between January 1, 1980 and December 31, 2010 in four European sites (Jokioinen, Dikopshof, Lleida and Lublin) from different climatic zones were modeled and forecasted. In our forecasting we used the methods of the Box-Jenkins and Holt- Winters seasonal auto regressive integrated moving-average, the autoregressive integrated moving-average with external regressors in the form of Fourier terms and the time series regression, including trend and seasonality components methodology with R software. It was demonstrated that obtained models are able to capture the dynamics of the time series data and to produce sensible forecasts.
An algorithm of Saxena-Easo on fuzzy time series forecasting
Ramadhani, L. C.; Anggraeni, D.; Kamsyakawuni, A.; Hadi, A. F.
2018-04-01
This paper presents a forecast model of Saxena-Easo fuzzy time series prediction to study the prediction of Indonesia inflation rate in 1970-2016. We use MATLAB software to compute this method. The algorithm of Saxena-Easo fuzzy time series doesn’t need stationarity like conventional forecasting method, capable of dealing with the value of time series which are linguistic and has the advantage of reducing the calculation, time and simplifying the calculation process. Generally it’s focus on percentage change as the universe discourse, interval partition and defuzzification. The result indicate that between the actual data and the forecast data are close enough with Root Mean Square Error (RMSE) = 1.5289.
Forecasting autoregressive time series under changing persistence
DEFF Research Database (Denmark)
Kruse, Robinson
Changing persistence in time series models means that a structural change from nonstationarity to stationarity or vice versa occurs over time. Such a change has important implications for forecasting, as negligence may lead to inaccurate model predictions. This paper derives generally applicable...
A four-stage hybrid model for hydrological time series forecasting.
Di, Chongli; Yang, Xiaohua; Wang, Xiaochao
2014-01-01
Hydrological time series forecasting remains a difficult task due to its complicated nonlinear, non-stationary and multi-scale characteristics. To solve this difficulty and improve the prediction accuracy, a novel four-stage hybrid model is proposed for hydrological time series forecasting based on the principle of 'denoising, decomposition and ensemble'. The proposed model has four stages, i.e., denoising, decomposition, components prediction and ensemble. In the denoising stage, the empirical mode decomposition (EMD) method is utilized to reduce the noises in the hydrological time series. Then, an improved method of EMD, the ensemble empirical mode decomposition (EEMD), is applied to decompose the denoised series into a number of intrinsic mode function (IMF) components and one residual component. Next, the radial basis function neural network (RBFNN) is adopted to predict the trend of all of the components obtained in the decomposition stage. In the final ensemble prediction stage, the forecasting results of all of the IMF and residual components obtained in the third stage are combined to generate the final prediction results, using a linear neural network (LNN) model. For illustration and verification, six hydrological cases with different characteristics are used to test the effectiveness of the proposed model. The proposed hybrid model performs better than conventional single models, the hybrid models without denoising or decomposition and the hybrid models based on other methods, such as the wavelet analysis (WA)-based hybrid models. In addition, the denoising and decomposition strategies decrease the complexity of the series and reduce the difficulties of the forecasting. With its effective denoising and accurate decomposition ability, high prediction precision and wide applicability, the new model is very promising for complex time series forecasting. This new forecast model is an extension of nonlinear prediction models.
A Four-Stage Hybrid Model for Hydrological Time Series Forecasting
Di, Chongli; Yang, Xiaohua; Wang, Xiaochao
2014-01-01
Hydrological time series forecasting remains a difficult task due to its complicated nonlinear, non-stationary and multi-scale characteristics. To solve this difficulty and improve the prediction accuracy, a novel four-stage hybrid model is proposed for hydrological time series forecasting based on the principle of ‘denoising, decomposition and ensemble’. The proposed model has four stages, i.e., denoising, decomposition, components prediction and ensemble. In the denoising stage, the empirical mode decomposition (EMD) method is utilized to reduce the noises in the hydrological time series. Then, an improved method of EMD, the ensemble empirical mode decomposition (EEMD), is applied to decompose the denoised series into a number of intrinsic mode function (IMF) components and one residual component. Next, the radial basis function neural network (RBFNN) is adopted to predict the trend of all of the components obtained in the decomposition stage. In the final ensemble prediction stage, the forecasting results of all of the IMF and residual components obtained in the third stage are combined to generate the final prediction results, using a linear neural network (LNN) model. For illustration and verification, six hydrological cases with different characteristics are used to test the effectiveness of the proposed model. The proposed hybrid model performs better than conventional single models, the hybrid models without denoising or decomposition and the hybrid models based on other methods, such as the wavelet analysis (WA)-based hybrid models. In addition, the denoising and decomposition strategies decrease the complexity of the series and reduce the difficulties of the forecasting. With its effective denoising and accurate decomposition ability, high prediction precision and wide applicability, the new model is very promising for complex time series forecasting. This new forecast model is an extension of nonlinear prediction models. PMID:25111782
New Models for Forecasting Enrollments: Fuzzy Time Series and Neural Network Approaches.
Song, Qiang; Chissom, Brad S.
Since university enrollment forecasting is very important, many different methods and models have been proposed by researchers. Two new methods for enrollment forecasting are introduced: (1) the fuzzy time series model; and (2) the artificial neural networks model. Fuzzy time series has been proposed to deal with forecasting problems within a…
A perturbative approach for enhancing the performance of time series forecasting.
de Mattos Neto, Paulo S G; Ferreira, Tiago A E; Lima, Aranildo R; Vasconcelos, Germano C; Cavalcanti, George D C
2017-04-01
This paper proposes a method to perform time series prediction based on perturbation theory. The approach is based on continuously adjusting an initial forecasting model to asymptotically approximate a desired time series model. First, a predictive model generates an initial forecasting for a time series. Second, a residual time series is calculated as the difference between the original time series and the initial forecasting. If that residual series is not white noise, then it can be used to improve the accuracy of the initial model and a new predictive model is adjusted using residual series. The whole process is repeated until convergence or the residual series becomes white noise. The output of the method is then given by summing up the outputs of all trained predictive models in a perturbative sense. To test the method, an experimental investigation was conducted on six real world time series. A comparison was made with six other methods experimented and ten other results found in the literature. Results show that not only the performance of the initial model is significantly improved but also the proposed method outperforms the other results previously published. Copyright © 2017 Elsevier Ltd. All rights reserved.
Conditional time series forecasting with convolutional neural networks
A. Borovykh (Anastasia); S.M. Bohte (Sander); C.W. Oosterlee (Cornelis)
2017-01-01
textabstractForecasting financial time series using past observations has been a significant topic of interest. While temporal relationships in the data exist, they are difficult to analyze and predict accurately due to the non-linear trends and noise present in the series. We propose to learn these
A hybrid approach EMD-HW for short-term forecasting of daily stock market time series data
Awajan, Ahmad Mohd; Ismail, Mohd Tahir
2017-08-01
Recently, forecasting time series has attracted considerable attention in the field of analyzing financial time series data, specifically within the stock market index. Moreover, stock market forecasting is a challenging area of financial time-series forecasting. In this study, a hybrid methodology between Empirical Mode Decomposition with the Holt-Winter method (EMD-HW) is used to improve forecasting performances in financial time series. The strength of this EMD-HW lies in its ability to forecast non-stationary and non-linear time series without a need to use any transformation method. Moreover, EMD-HW has a relatively high accuracy and offers a new forecasting method in time series. The daily stock market time series data of 11 countries is applied to show the forecasting performance of the proposed EMD-HW. Based on the three forecast accuracy measures, the results indicate that EMD-HW forecasting performance is superior to traditional Holt-Winter forecasting method.
Markov Chain Modelling for Short-Term NDVI Time Series Forecasting
Directory of Open Access Journals (Sweden)
Stepčenko Artūrs
2016-12-01
Full Text Available In this paper, the NDVI time series forecasting model has been developed based on the use of discrete time, continuous state Markov chain of suitable order. The normalised difference vegetation index (NDVI is an indicator that describes the amount of chlorophyll (the green mass and shows the relative density and health of vegetation; therefore, it is an important variable for vegetation forecasting. A Markov chain is a stochastic process that consists of a state space. This stochastic process undergoes transitions from one state to another in the state space with some probabilities. A Markov chain forecast model is flexible in accommodating various forecast assumptions and structures. The present paper discusses the considerations and techniques in building a Markov chain forecast model at each step. Continuous state Markov chain model is analytically described. Finally, the application of the proposed Markov chain model is illustrated with reference to a set of NDVI time series data.
A Time Series Forecasting Method
Directory of Open Access Journals (Sweden)
Wang Zhao-Yu
2017-01-01
Full Text Available This paper proposes a novel time series forecasting method based on a weighted self-constructing clustering technique. The weighted self-constructing clustering processes all the data patterns incrementally. If a data pattern is not similar enough to an existing cluster, it forms a new cluster of its own. However, if a data pattern is similar enough to an existing cluster, it is removed from the cluster it currently belongs to and added to the most similar cluster. During the clustering process, weights are learned for each cluster. Given a series of time-stamped data up to time t, we divide it into a set of training patterns. By using the weighted self-constructing clustering, the training patterns are grouped into a set of clusters. To estimate the value at time t + 1, we find the k nearest neighbors of the input pattern and use these k neighbors to decide the estimation. Experimental results are shown to demonstrate the effectiveness of the proposed approach.
Time series forecasting based on deep extreme learning machine
Guo, Xuqi; Pang, Y.; Yan, Gaowei; Qiao, Tiezhu; Yang, Guang-Hong; Yang, Dan
2017-01-01
Multi-layer Artificial Neural Networks (ANN) has caught widespread attention as a new method for time series forecasting due to the ability of approximating any nonlinear function. In this paper, a new local time series prediction model is established with the nearest neighbor domain theory, in
Ridge Polynomial Neural Network with Error Feedback for Time Series Forecasting.
Waheeb, Waddah; Ghazali, Rozaida; Herawan, Tutut
2016-01-01
Time series forecasting has gained much attention due to its many practical applications. Higher-order neural network with recurrent feedback is a powerful technique that has been used successfully for time series forecasting. It maintains fast learning and the ability to learn the dynamics of the time series over time. Network output feedback is the most common recurrent feedback for many recurrent neural network models. However, not much attention has been paid to the use of network error feedback instead of network output feedback. In this study, we propose a novel model, called Ridge Polynomial Neural Network with Error Feedback (RPNN-EF) that incorporates higher order terms, recurrence and error feedback. To evaluate the performance of RPNN-EF, we used four univariate time series with different forecasting horizons, namely star brightness, monthly smoothed sunspot numbers, daily Euro/Dollar exchange rate, and Mackey-Glass time-delay differential equation. We compared the forecasting performance of RPNN-EF with the ordinary Ridge Polynomial Neural Network (RPNN) and the Dynamic Ridge Polynomial Neural Network (DRPNN). Simulation results showed an average 23.34% improvement in Root Mean Square Error (RMSE) with respect to RPNN and an average 10.74% improvement with respect to DRPNN. That means that using network errors during training helps enhance the overall forecasting performance for the network.
Ridge Polynomial Neural Network with Error Feedback for Time Series Forecasting.
Directory of Open Access Journals (Sweden)
Waddah Waheeb
Full Text Available Time series forecasting has gained much attention due to its many practical applications. Higher-order neural network with recurrent feedback is a powerful technique that has been used successfully for time series forecasting. It maintains fast learning and the ability to learn the dynamics of the time series over time. Network output feedback is the most common recurrent feedback for many recurrent neural network models. However, not much attention has been paid to the use of network error feedback instead of network output feedback. In this study, we propose a novel model, called Ridge Polynomial Neural Network with Error Feedback (RPNN-EF that incorporates higher order terms, recurrence and error feedback. To evaluate the performance of RPNN-EF, we used four univariate time series with different forecasting horizons, namely star brightness, monthly smoothed sunspot numbers, daily Euro/Dollar exchange rate, and Mackey-Glass time-delay differential equation. We compared the forecasting performance of RPNN-EF with the ordinary Ridge Polynomial Neural Network (RPNN and the Dynamic Ridge Polynomial Neural Network (DRPNN. Simulation results showed an average 23.34% improvement in Root Mean Square Error (RMSE with respect to RPNN and an average 10.74% improvement with respect to DRPNN. That means that using network errors during training helps enhance the overall forecasting performance for the network.
Ratio-based lengths of intervals to improve fuzzy time series forecasting.
Huarng, Kunhuang; Yu, Tiffany Hui-Kuang
2006-04-01
The objective of this study is to explore ways of determining the useful lengths of intervals in fuzzy time series. It is suggested that ratios, instead of equal lengths of intervals, can more properly represent the intervals among observations. Ratio-based lengths of intervals are, therefore, proposed to improve fuzzy time series forecasting. Algebraic growth data, such as enrollments and the stock index, and exponential growth data, such as inventory demand, are chosen as the forecasting targets, before forecasting based on the various lengths of intervals is performed. Furthermore, sensitivity analyses are also carried out for various percentiles. The ratio-based lengths of intervals are found to outperform the effective lengths of intervals, as well as the arbitrary ones in regard to the different statistical measures. The empirical analysis suggests that the ratio-based lengths of intervals can also be used to improve fuzzy time series forecasting.
Medina, Daniel C; Findley, Sally E; Guindo, Boubacar; Doumbia, Seydou
2007-11-21
Much of the developing world, particularly sub-Saharan Africa, exhibits high levels of morbidity and mortality associated with diarrhea, acute respiratory infection, and malaria. With the increasing awareness that the aforementioned infectious diseases impose an enormous burden on developing countries, public health programs therein could benefit from parsimonious general-purpose forecasting methods to enhance infectious disease intervention. Unfortunately, these disease time-series often i) suffer from non-stationarity; ii) exhibit large inter-annual plus seasonal fluctuations; and, iii) require disease-specific tailoring of forecasting methods. In this longitudinal retrospective (01/1996-06/2004) investigation, diarrhea, acute respiratory infection of the lower tract, and malaria consultation time-series are fitted with a general-purpose econometric method, namely the multiplicative Holt-Winters, to produce contemporaneous on-line forecasts for the district of Niono, Mali. This method accommodates seasonal, as well as inter-annual, fluctuations and produces reasonably accurate median 2- and 3-month horizon forecasts for these non-stationary time-series, i.e., 92% of the 24 time-series forecasts generated (2 forecast horizons, 3 diseases, and 4 age categories = 24 time-series forecasts) have mean absolute percentage errors circa 25%. The multiplicative Holt-Winters forecasting method: i) performs well across diseases with dramatically distinct transmission modes and hence it is a strong general-purpose forecasting method candidate for non-stationary epidemiological time-series; ii) obliquely captures prior non-linear interactions between climate and the aforementioned disease dynamics thus, obviating the need for more complex disease-specific climate-based parametric forecasting methods in the district of Niono; furthermore, iii) readily decomposes time-series into seasonal components thereby potentially assisting with programming of public health interventions
Time series modelling and forecasting of emergency department overcrowding.
Kadri, Farid; Harrou, Fouzi; Chaabane, Sondès; Tahon, Christian
2014-09-01
Efficient management of patient flow (demand) in emergency departments (EDs) has become an urgent issue for many hospital administrations. Today, more and more attention is being paid to hospital management systems to optimally manage patient flow and to improve management strategies, efficiency and safety in such establishments. To this end, EDs require significant human and material resources, but unfortunately these are limited. Within such a framework, the ability to accurately forecast demand in emergency departments has considerable implications for hospitals to improve resource allocation and strategic planning. The aim of this study was to develop models for forecasting daily attendances at the hospital emergency department in Lille, France. The study demonstrates how time-series analysis can be used to forecast, at least in the short term, demand for emergency services in a hospital emergency department. The forecasts were based on daily patient attendances at the paediatric emergency department in Lille regional hospital centre, France, from January 2012 to December 2012. An autoregressive integrated moving average (ARIMA) method was applied separately to each of the two GEMSA categories and total patient attendances. Time-series analysis was shown to provide a useful, readily available tool for forecasting emergency department demand.
Stacked Heterogeneous Neural Networks for Time Series Forecasting
Directory of Open Access Journals (Sweden)
Florin Leon
2010-01-01
Full Text Available A hybrid model for time series forecasting is proposed. It is a stacked neural network, containing one normal multilayer perceptron with bipolar sigmoid activation functions, and the other with an exponential activation function in the output layer. As shown by the case studies, the proposed stacked hybrid neural model performs well on a variety of benchmark time series. The combination of weights of the two stack components that leads to optimal performance is also studied.
Forecasting Inflation Using Interest-Rate and Time-Series Models: Some International Evidence.
Hafer, R W; Hein, Scott E
1990-01-01
It has been suggested that inflation forecasts derived from short-term interest rates are as accurate as time-series forecasts. Previous analyses of this notion have focused on U.S. data, providing mixed results. In this article, the authors extend previous work by testing the hypothesis using data taken from the United States and five other countries. Using monthly Eurocurrency rates and the consumer price index for the period 1967-86, their results indicate that time-series forecasts of inf...
Forecasting the Reference Evapotranspiration Using Time Series Model
Directory of Open Access Journals (Sweden)
H. Zare Abyaneh
2016-10-01
Full Text Available Introduction: Reference evapotranspiration is one of the most important factors in irrigation timing and field management. Moreover, reference evapotranspiration forecasting can play a vital role in future developments. Therefore in this study, the seasonal autoregressive integrated moving average (ARIMA model was used to forecast the reference evapotranspiration time series in the Esfahan, Semnan, Shiraz, Kerman, and Yazd synoptic stations. Materials and Methods: In the present study in all stations (characteristics of the synoptic stations are given in Table 1, the meteorological data, including mean, maximum and minimum air temperature, relative humidity, dry-and wet-bulb temperature, dew-point temperature, wind speed, precipitation, air vapor pressure and sunshine hours were collected from the Islamic Republic of Iran Meteorological Organization (IRIMO for the 41 years from 1965 to 2005. The FAO Penman-Monteith equation was used to calculate the monthly reference evapotranspiration in the five synoptic stations and the evapotranspiration time series were formed. The unit root test was used to identify whether the time series was stationary, then using the Box-Jenkins method, seasonal ARIMA models were applied to the sample data. Table 1. The geographical location and climate conditions of the synoptic stations Station\tGeographical location\tAltitude (m\tMean air temperature (°C\tMean precipitation (mm\tClimate, according to the De Martonne index classification Longitude (E\tLatitude (N Annual\tMin. and Max. Esfahan\t51° 40'\t32° 37'\t1550.4\t16.36\t9.4-23.3\t122\tArid Semnan\t53° 33'\t35° 35'\t1130.8\t18.0\t12.4-23.8\t140\tArid Shiraz\t52° 36'\t29° 32'\t1484\t18.0\t10.2-25.9\t324\tSemi-arid Kerman\t56° 58'\t30° 15'\t1753.8\t15.6\t6.7-24.6\t142\tArid Yazd\t54° 17'\t31° 54'\t1237.2\t19.2\t11.8-26.0\t61\tArid Results and Discussion: The monthly meteorological data were used as input for the Ref-ET software and monthly reference
Seasonal time series forecasting: a comparative study of arima and ...
African Journals Online (AJOL)
This paper addresses the concerns of Faraway and Chatfield (1998) who questioned the forecasting ability of Artificial Neural Networks (ANN). In particular the paper compares the performance of Artificial Neural Networks (ANN) and ARIMA models in forecasting of seasonal (monthly) Time series. Using the Airline data ...
Trend time-series modeling and forecasting with neural networks.
Qi, Min; Zhang, G Peter
2008-05-01
Despite its great importance, there has been no general consensus on how to model the trends in time-series data. Compared to traditional approaches, neural networks (NNs) have shown some promise in time-series forecasting. This paper investigates how to best model trend time series using NNs. Four different strategies (raw data, raw data with time index, detrending, and differencing) are used to model various trend patterns (linear, nonlinear, deterministic, stochastic, and breaking trend). We find that with NNs differencing often gives meritorious results regardless of the underlying data generating processes (DGPs). This finding is also confirmed by the real gross national product (GNP) series.
Intuitionistic Fuzzy Time Series Forecasting Model Based on Intuitionistic Fuzzy Reasoning
Directory of Open Access Journals (Sweden)
Ya’nan Wang
2016-01-01
Full Text Available Fuzzy sets theory cannot describe the data comprehensively, which has greatly limited the objectivity of fuzzy time series in uncertain data forecasting. In this regard, an intuitionistic fuzzy time series forecasting model is built. In the new model, a fuzzy clustering algorithm is used to divide the universe of discourse into unequal intervals, and a more objective technique for ascertaining the membership function and nonmembership function of the intuitionistic fuzzy set is proposed. On these bases, forecast rules based on intuitionistic fuzzy approximate reasoning are established. At last, contrast experiments on the enrollments of the University of Alabama and the Taiwan Stock Exchange Capitalization Weighted Stock Index are carried out. The results show that the new model has a clear advantage of improving the forecast accuracy.
A stochastic HMM-based forecasting model for fuzzy time series.
Li, Sheng-Tun; Cheng, Yi-Chung
2010-10-01
Recently, fuzzy time series have attracted more academic attention than traditional time series due to their capability of dealing with the uncertainty and vagueness inherent in the data collected. The formulation of fuzzy relations is one of the key issues affecting forecasting results. Most of the present works adopt IF-THEN rules for relationship representation, which leads to higher computational overhead and rule redundancy. Sullivan and Woodall proposed a Markov-based formulation and a forecasting model to reduce computational overhead; however, its applicability is limited to handling one-factor problems. In this paper, we propose a novel forecasting model based on the hidden Markov model by enhancing Sullivan and Woodall's work to allow handling of two-factor forecasting problems. Moreover, in order to make the nature of conjecture and randomness of forecasting more realistic, the Monte Carlo method is adopted to estimate the outcome. To test the effectiveness of the resulting stochastic model, we conduct two experiments and compare the results with those from other models. The first experiment consists of forecasting the daily average temperature and cloud density in Taipei, Taiwan, and the second experiment is based on the Taiwan Weighted Stock Index by forecasting the exchange rate of the New Taiwan dollar against the U.S. dollar. In addition to improving forecasting accuracy, the proposed model adheres to the central limit theorem, and thus, the result statistically approximates to the real mean of the target value being forecast.
Hybrid model for forecasting time series with trend, seasonal and salendar variation patterns
Suhartono; Rahayu, S. P.; Prastyo, D. D.; Wijayanti, D. G. P.; Juliyanto
2017-09-01
Most of the monthly time series data in economics and business in Indonesia and other Moslem countries not only contain trend and seasonal, but also affected by two types of calendar variation effects, i.e. the effect of the number of working days or trading and holiday effects. The purpose of this research is to develop a hybrid model or a combination of several forecasting models to predict time series that contain trend, seasonal and calendar variation patterns. This hybrid model is a combination of classical models (namely time series regression and ARIMA model) and/or modern methods (artificial intelligence method, i.e. Artificial Neural Networks). A simulation study was used to show that the proposed procedure for building the hybrid model could work well for forecasting time series with trend, seasonal and calendar variation patterns. Furthermore, the proposed hybrid model is applied for forecasting real data, i.e. monthly data about inflow and outflow of currency at Bank Indonesia. The results show that the hybrid model tend to provide more accurate forecasts than individual forecasting models. Moreover, this result is also in line with the third results of the M3 competition, i.e. the hybrid model on average provides a more accurate forecast than the individual model.
Kriging Methodology and Its Development in Forecasting Econometric Time Series
Directory of Open Access Journals (Sweden)
Andrej Gajdoš
2017-03-01
Full Text Available One of the approaches for forecasting future values of a time series or unknown spatial data is kriging. The main objective of the paper is to introduce a general scheme of kriging in forecasting econometric time series using a family of linear regression time series models (shortly named as FDSLRM which apply regression not only to a trend but also to a random component of the observed time series. Simultaneously performing a Monte Carlo simulation study with a real electricity consumption dataset in the R computational langure and environment, we investigate the well-known problem of “negative” estimates of variance components when kriging predictions fail. Our following theoretical analysis, including also the modern apparatus of advanced multivariate statistics, gives us the formulation and proof of a general theorem about the explicit form of moments (up to sixth order for a Gaussian time series observation. This result provides a basis for further theoretical and computational research in the kriging methodology development.
Forecasting long memory time series under a break in persistence
DEFF Research Database (Denmark)
Heinen, Florian; Sibbertsen, Philipp; Kruse, Robinson
We consider the problem of forecasting time series with long memory when the memory parameter is subject to a structural break. By means of a large-scale Monte Carlo study we show that ignoring such a change in persistence leads to substantially reduced forecasting precision. The strength...... of this effect depends on whether the memory parameter is increasing or decreasing over time. A comparison of six forecasting strategies allows us to conclude that pre-testing for a change in persistence is highly recommendable in our setting. In addition we provide an empirical example which underlines...
Artificial neural networks applied to forecasting time series.
Montaño Moreno, Juan J; Palmer Pol, Alfonso; Muñoz Gracia, Pilar
2011-04-01
This study offers a description and comparison of the main models of Artificial Neural Networks (ANN) which have proved to be useful in time series forecasting, and also a standard procedure for the practical application of ANN in this type of task. The Multilayer Perceptron (MLP), Radial Base Function (RBF), Generalized Regression Neural Network (GRNN), and Recurrent Neural Network (RNN) models are analyzed. With this aim in mind, we use a time series made up of 244 time points. A comparative study establishes that the error made by the four neural network models analyzed is less than 10%. In accordance with the interpretation criteria of this performance, it can be concluded that the neural network models show a close fit regarding their forecasting capacity. The model with the best performance is the RBF, followed by the RNN and MLP. The GRNN model is the one with the worst performance. Finally, we analyze the advantages and limitations of ANN, the possible solutions to these limitations, and provide an orientation towards future research.
forecasting with nonlinear time series model: a monte-carlo
African Journals Online (AJOL)
PUBLICATIONS1
Carlo method of forecasting using a special nonlinear time series model, called logistic smooth transition ... We illustrate this new method using some simulation ..... in MATLAB 7.5.0. ... process (DGP) using the logistic smooth transi-.
Hamid, Nor Zila Abd; Adenan, Nur Hamiza; Noorani, Mohd Salmi Md
2017-08-01
Forecasting and analyzing the ozone (O3) concentration time series is important because the pollutant is harmful to health. This study is a pilot study for forecasting and analyzing the O3 time series in one of Malaysian educational area namely Shah Alam using chaotic approach. Through this approach, the observed hourly scalar time series is reconstructed into a multi-dimensional phase space, which is then used to forecast the future time series through the local linear approximation method. The main purpose is to forecast the high O3 concentrations. The original method performed poorly but the improved method addressed the weakness thereby enabling the high concentrations to be successfully forecast. The correlation coefficient between the observed and forecasted time series through the improved method is 0.9159 and both the mean absolute error and root mean squared error are low. Thus, the improved method is advantageous. The time series analysis by means of the phase space plot and Cao method identified the presence of low-dimensional chaotic dynamics in the observed O3 time series. Results showed that at least seven factors affect the studied O3 time series, which is consistent with the listed factors from the diurnal variations investigation and the sensitivity analysis from past studies. In conclusion, chaotic approach has been successfully forecast and analyzes the O3 time series in educational area of Shah Alam. These findings are expected to help stakeholders such as Ministry of Education and Department of Environment in having a better air pollution management.
Forecasting air quality time series using deep learning.
Freeman, Brian S; Taylor, Graham; Gharabaghi, Bahram; Thé, Jesse
2018-04-13
This paper presents one of the first applications of deep learning (DL) techniques to predict air pollution time series. Air quality management relies extensively on time series data captured at air monitoring stations as the basis of identifying population exposure to airborne pollutants and determining compliance with local ambient air standards. In this paper, 8 hr averaged surface ozone (O 3 ) concentrations were predicted using deep learning consisting of a recurrent neural network (RNN) with long short-term memory (LSTM). Hourly air quality and meteorological data were used to train and forecast values up to 72 hours with low error rates. The LSTM was able to forecast the duration of continuous O 3 exceedances as well. Prior to training the network, the dataset was reviewed for missing data and outliers. Missing data were imputed using a novel technique that averaged gaps less than eight time steps with incremental steps based on first-order differences of neighboring time periods. Data were then used to train decision trees to evaluate input feature importance over different time prediction horizons. The number of features used to train the LSTM model was reduced from 25 features to 5 features, resulting in improved accuracy as measured by Mean Absolute Error (MAE). Parameter sensitivity analysis identified look-back nodes associated with the RNN proved to be a significant source of error if not aligned with the prediction horizon. Overall, MAE's less than 2 were calculated for predictions out to 72 hours. Novel deep learning techniques were used to train an 8-hour averaged ozone forecast model. Missing data and outliers within the captured data set were replaced using a new imputation method that generated calculated values closer to the expected value based on the time and season. Decision trees were used to identify input variables with the greatest importance. The methods presented in this paper allow air managers to forecast long range air pollution
A Bias and Variance Analysis for Multistep-Ahead Time Series Forecasting.
Ben Taieb, Souhaib; Atiya, Amir F
2016-01-01
Multistep-ahead forecasts can either be produced recursively by iterating a one-step-ahead time series model or directly by estimating a separate model for each forecast horizon. In addition, there are other strategies; some of them combine aspects of both aforementioned concepts. In this paper, we present a comprehensive investigation into the bias and variance behavior of multistep-ahead forecasting strategies. We provide a detailed review of the different multistep-ahead strategies. Subsequently, we perform a theoretical study that derives the bias and variance for a number of forecasting strategies. Finally, we conduct a Monte Carlo experimental study that compares and evaluates the bias and variance performance of the different strategies. From the theoretical and the simulation studies, we analyze the effect of different factors, such as the forecast horizon and the time series length, on the bias and variance components, and on the different multistep-ahead strategies. Several lessons are learned, and recommendations are given concerning the advantages, disadvantages, and best conditions of use of each strategy.
Forecasting with quantitative methods the impact of special events in time series
Nikolopoulos, Konstantinos
2010-01-01
Abstract Quantitative methods are very successful for producing baseline forecasts of time series; however these models fail to forecast neither the timing nor the impact of special events such as promotions or strikes. In most of the cases the timing of such events is not known so they are usually referred as shocks (economics) or special events (forecasting). Sometimes the timing of such events is known a priori (i.e. a future promotion); but even then the impact of the forthcom...
Ningrum, R. W.; Surarso, B.; Farikhin; Safarudin, Y. M.
2018-03-01
This paper proposes the combination of Firefly Algorithm (FA) and Chen Fuzzy Time Series Forecasting. Most of the existing fuzzy forecasting methods based on fuzzy time series use the static length of intervals. Therefore, we apply an artificial intelligence, i.e., Firefly Algorithm (FA) to set non-stationary length of intervals for each cluster on Chen Method. The method is evaluated by applying on the Jakarta Composite Index (IHSG) and compare with classical Chen Fuzzy Time Series Forecasting. Its performance verified through simulation using Matlab.
Papacharalampous, Georgia; Tyralis, Hristos; Koutsoyiannis, Demetris
2018-02-01
We investigate the predictability of monthly temperature and precipitation by applying automatic univariate time series forecasting methods to a sample of 985 40-year-long monthly temperature and 1552 40-year-long monthly precipitation time series. The methods include a naïve one based on the monthly values of the last year, as well as the random walk (with drift), AutoRegressive Fractionally Integrated Moving Average (ARFIMA), exponential smoothing state-space model with Box-Cox transformation, ARMA errors, Trend and Seasonal components (BATS), simple exponential smoothing, Theta and Prophet methods. Prophet is a recently introduced model inspired by the nature of time series forecasted at Facebook and has not been applied to hydrometeorological time series before, while the use of random walk, BATS, simple exponential smoothing and Theta is rare in hydrology. The methods are tested in performing multi-step ahead forecasts for the last 48 months of the data. We further investigate how different choices of handling the seasonality and non-normality affect the performance of the models. The results indicate that: (a) all the examined methods apart from the naïve and random walk ones are accurate enough to be used in long-term applications; (b) monthly temperature and precipitation can be forecasted to a level of accuracy which can barely be improved using other methods; (c) the externally applied classical seasonal decomposition results mostly in better forecasts compared to the automatic seasonal decomposition used by the BATS and Prophet methods; and (d) Prophet is competitive, especially when it is combined with externally applied classical seasonal decomposition.
Forecasting electricity spot-prices using linear univariate time-series models
International Nuclear Information System (INIS)
Cuaresma, Jesus Crespo; Hlouskova, Jaroslava; Kossmeier, Stephan; Obersteiner, Michael
2004-01-01
This paper studies the forecasting abilities of a battery of univariate models on hourly electricity spot prices, using data from the Leipzig Power Exchange. The specifications studied include autoregressive models, autoregressive-moving average models and unobserved component models. The results show that specifications, where each hour of the day is modelled separately present uniformly better forecasting properties than specifications for the whole time-series, and that the inclusion of simple probabilistic processes for the arrival of extreme price events can lead to improvements in the forecasting abilities of univariate models for electricity spot prices. (Author)
A Time-Series Water Level Forecasting Model Based on Imputation and Variable Selection Method
Jun-He Yang; Ching-Hsue Cheng; Chia-Pan Chan
2017-01-01
Reservoirs are important for households and impact the national economy. This paper proposed a time-series forecasting model based on estimating a missing value followed by variable selection to forecast the reservoir's water level. This study collected data from the Taiwan Shimen Reservoir as well as daily atmospheric data from 2008 to 2015. The two datasets are concatenated into an integrated dataset based on ordering of the data as a research dataset. The proposed time-series forecasting m...
Year Ahead Demand Forecast of City Natural Gas Using Seasonal Time Series Methods
Directory of Open Access Journals (Sweden)
Mustafa Akpinar
2016-09-01
Full Text Available Consumption of natural gas, a major clean energy source, increases as energy demand increases. We studied specifically the Turkish natural gas market. Turkey’s natural gas consumption increased as well in parallel with the world‘s over the last decade. This consumption growth in Turkey has led to the formation of a market structure for the natural gas industry. This significant increase requires additional investments since a rise in consumption capacity is expected. One of the reasons for the consumption increase is the user-based natural gas consumption influence. This effect yields imbalances in demand forecasts and if the error rates are out of bounds, penalties may occur. In this paper, three univariate statistical methods, which have not been previously investigated for mid-term year-ahead monthly natural gas forecasting, are used to forecast natural gas demand in Turkey’s Sakarya province. Residential and low-consumption commercial data is used, which may contain seasonality. The goal of this paper is minimizing more or less gas tractions on mid-term consumption while improving the accuracy of demand forecasting. In forecasting models, seasonality and single variable impacts reinforce forecasts. This paper studies time series decomposition, Holt-Winters exponential smoothing and autoregressive integrated moving average (ARIMA methods. Here, 2011–2014 monthly data were prepared and divided into two series. The first series is 2011–2013 monthly data used for finding seasonal effects and model requirements. The second series is 2014 monthly data used for forecasting. For the ARIMA method, a stationary series was prepared and transformation process prior to forecasting was done. Forecasting results confirmed that as the computation complexity of the model increases, forecasting accuracy increases with lower error rates. Also, forecasting errors and the coefficients of determination values give more consistent results. Consequently
Hansen, J V; Nelson, R D
1997-01-01
Ever since the initial planning for the 1997 Utah legislative session, neural-network forecasting techniques have provided valuable insights for analysts forecasting tax revenues. These revenue estimates are critically important since agency budgets, support for education, and improvements to infrastructure all depend on their accuracy. Underforecasting generates windfalls that concern taxpayers, whereas overforecasting produces budget shortfalls that cause inadequately funded commitments. The pattern finding ability of neural networks gives insightful and alternative views of the seasonal and cyclical components commonly found in economic time series data. Two applications of neural networks to revenue forecasting clearly demonstrate how these models complement traditional time series techniques. In the first, preoccupation with a potential downturn in the economy distracts analysis based on traditional time series methods so that it overlooks an emerging new phenomenon in the data. In this case, neural networks identify the new pattern that then allows modification of the time series models and finally gives more accurate forecasts. In the second application, data structure found by traditional statistical tools allows analysts to provide neural networks with important information that the networks then use to create more accurate models. In summary, for the Utah revenue outlook, the insights that result from a portfolio of forecasts that includes neural networks exceeds the understanding generated from strictly statistical forecasting techniques. In this case, the synergy clearly results in the whole of the portfolio of forecasts being more accurate than the sum of the individual parts.
A comparison of the stochastic and machine learning approaches in hydrologic time series forecasting
Kim, T.; Joo, K.; Seo, J.; Heo, J. H.
2016-12-01
Hydrologic time series forecasting is an essential task in water resources management and it becomes more difficult due to the complexity of runoff process. Traditional stochastic models such as ARIMA family has been used as a standard approach in time series modeling and forecasting of hydrological variables. Due to the nonlinearity in hydrologic time series data, machine learning approaches has been studied with the advantage of discovering relevant features in a nonlinear relation among variables. This study aims to compare the predictability between the traditional stochastic model and the machine learning approach. Seasonal ARIMA model was used as the traditional time series model, and Random Forest model which consists of decision tree and ensemble method using multiple predictor approach was applied as the machine learning approach. In the application, monthly inflow data from 1986 to 2015 of Chungju dam in South Korea were used for modeling and forecasting. In order to evaluate the performances of the used models, one step ahead and multi-step ahead forecasting was applied. Root mean squared error and mean absolute error of two models were compared.
Synthetic river flow time series generator for dispatch and spot price forecast
International Nuclear Information System (INIS)
Flores, R.A.
2007-01-01
Decision-making in electricity markets is complicated by uncertainties in demand growth, power supplies and fuel prices. In Peru, where the electrical power system is highly dependent on water resources at dams and river flows, hydrological uncertainties play a primary role in planning, price and dispatch forecast. This paper proposed a signal processing method for generating new synthetic river flow time series as a support for planning and spot market price forecasting. River flow time series are natural phenomena representing a continuous-time domain process. As an alternative synthetic representation of the original river flow time series, this proposed signal processing method preserves correlations, basic statistics and seasonality. It takes into account deterministic, periodic and non periodic components such as those due to the El Nino Southern Oscillation phenomenon. The new synthetic time series has many correlations with the original river flow time series, rendering it suitable for possible replacement of the classical method of sorting historical river flow time series. As a dispatch and planning approach to spot pricing, the proposed method offers higher accuracy modeling by decomposing the signal into deterministic, periodic, non periodic and stochastic sub signals. 4 refs., 4 tabs., 13 figs
State-space forecasting of Schistosoma haematobium time-series in Niono, Mali.
Medina, Daniel C; Findley, Sally E; Doumbia, Seydou
2008-08-13
Much of the developing world, particularly sub-Saharan Africa, exhibits high levels of morbidity and mortality associated with infectious diseases. The incidence of Schistosoma sp.-which are neglected tropical diseases exposing and infecting more than 500 and 200 million individuals in 77 countries, respectively-is rising because of 1) numerous irrigation and hydro-electric projects, 2) steady shifts from nomadic to sedentary existence, and 3) ineffective control programs. Notwithstanding the colossal scope of these parasitic infections, less than 0.5% of Schistosoma sp. investigations have attempted to predict their spatial and or temporal distributions. Undoubtedly, public health programs in developing countries could benefit from parsimonious forecasting and early warning systems to enhance management of these parasitic diseases. In this longitudinal retrospective (01/1996-06/2004) investigation, the Schistosoma haematobium time-series for the district of Niono, Mali, was fitted with general-purpose exponential smoothing methods to generate contemporaneous on-line forecasts. These methods, which are encapsulated within a state-space framework, accommodate seasonal and inter-annual time-series fluctuations. Mean absolute percentage error values were circa 25% for 1- to 5-month horizon forecasts. The exponential smoothing state-space framework employed herein produced reasonably accurate forecasts for this time-series, which reflects the incidence of S. haematobium-induced terminal hematuria. It obliquely captured prior non-linear interactions between disease dynamics and exogenous covariates (e.g., climate, irrigation, and public health interventions), thus obviating the need for more complex forecasting methods in the district of Niono, Mali. Therefore, this framework could assist with managing and assessing S. haematobium transmission and intervention impact, respectively, in this district and potentially elsewhere in the Sahel.
Directory of Open Access Journals (Sweden)
Suhartono Suhartono
2005-01-01
Full Text Available Many business and economic time series are non-stationary time series that contain trend and seasonal variations. Seasonality is a periodic and recurrent pattern caused by factors such as weather, holidays, or repeating promotions. A stochastic trend is often accompanied with the seasonal variations and can have a significant impact on various forecasting methods. In this paper, we will investigate and compare some forecasting methods for modeling time series with both trend and seasonal patterns. These methods are Winter's, Decomposition, Time Series Regression, ARIMA and Neural Networks models. In this empirical research, we study on the effectiveness of the forecasting performance, particularly to answer whether a complex method always give a better forecast than a simpler method. We use a real data, that is airline passenger data. The result shows that the more complex model does not always yield a better result than a simpler one. Additionally, we also find the possibility to do further research especially the use of hybrid model by combining some forecasting method to get better forecast, for example combination between decomposition (as data preprocessing and neural network model.
Recurrent Neural Network For Forecasting Time Series With Long Memory Pattern
Walid; Alamsyah
2017-04-01
Recurrent Neural Network as one of the hybrid models are often used to predict and estimate the issues related to electricity, can be used to describe the cause of the swelling of electrical load which experienced by PLN. In this research will be developed RNN forecasting procedures at the time series with long memory patterns. Considering the application is the national electrical load which of course has a different trend with the condition of the electrical load in any country. This research produces the algorithm of time series forecasting which has long memory pattern using E-RNN after this referred to the algorithm of integrated fractional recurrent neural networks (FIRNN).The prediction results of long memory time series using models Fractional Integrated Recurrent Neural Network (FIRNN) showed that the model with the selection of data difference in the range of [-1,1] and the model of Fractional Integrated Recurrent Neural Network (FIRNN) (24,6,1) provides the smallest MSE value, which is 0.00149684.
Forecast models for suicide: Time-series analysis with data from Italy.
Preti, Antonio; Lentini, Gianluca
2016-01-01
The prediction of suicidal behavior is a complex task. To fine-tune targeted preventative interventions, predictive analytics (i.e. forecasting future risk of suicide) is more important than exploratory data analysis (pattern recognition, e.g. detection of seasonality in suicide time series). This study sets out to investigate the accuracy of forecasting models of suicide for men and women. A total of 101 499 male suicides and of 39 681 female suicides - occurred in Italy from 1969 to 2003 - were investigated. In order to apply the forecasting model and test its accuracy, the time series were split into a training set (1969 to 1996; 336 months) and a test set (1997 to 2003; 84 months). The main outcome was the accuracy of forecasting models on the monthly number of suicides. These measures of accuracy were used: mean absolute error; root mean squared error; mean absolute percentage error; mean absolute scaled error. In both male and female suicides a change in the trend pattern was observed, with an increase from 1969 onwards to reach a maximum around 1990 and decrease thereafter. The variances attributable to the seasonal and trend components were, respectively, 24% and 64% in male suicides, and 28% and 41% in female ones. Both annual and seasonal historical trends of monthly data contributed to forecast future trends of suicide with a margin of error around 10%. The finding is clearer in male than in female time series of suicide. The main conclusion of the study is that models taking seasonality into account seem to be able to derive information on deviation from the mean when this occurs as a zenith, but they fail to reproduce it when it occurs as a nadir. Preventative efforts should concentrate on the factors that influence the occurrence of increases above the main trend in both seasonal and cyclic patterns of suicides.
Directory of Open Access Journals (Sweden)
Levi Lopes Teixeira
2015-12-01
Full Text Available Time series forecasting is widely used in various areas of human knowledge, especially in the planning and strategic direction of companies. The success of this task depends on the forecasting techniques applied. In this paper, a hybrid approach to project time series is suggested. To validate the methodology, a time series already modeled by other authors was chosen, allowing the comparison of results. The proposed methodology includes the following techniques: wavelet shrinkage, wavelet decomposition at level r, and artificial neural networks (ANN. Firstly, a time series to be forecasted is submitted to the proposed wavelet filtering method, which decomposes it to components of trend and linear residue. Then, both are decomposed via level r wavelet decomposition, generating r + 1 Wavelet Components (WCs for each one; and then each WC is individually modeled by an ANN. Finally, the predictions for all WCs are linearly combined, producing forecasts to the underlying time series. For evaluating purposes, the time series of Canadian Lynx has been used, and all results achieved by the proposed method were better than others in existing literature.
Rasim; Junaeti, E.; Wirantika, R.
2018-01-01
Accurate forecasting for the sale of a product depends on the forecasting method used. The purpose of this research is to build motorcycle sales forecasting application using Fuzzy Time Series method combined with interval determination using automatic clustering algorithm. Forecasting is done using the sales data of motorcycle sales in the last ten years. Then the error rate of forecasting is measured using Means Percentage Error (MPE) and Means Absolute Percentage Error (MAPE). The results of forecasting in the one-year period obtained in this study are included in good accuracy.
A Time-Series Water Level Forecasting Model Based on Imputation and Variable Selection Method.
Yang, Jun-He; Cheng, Ching-Hsue; Chan, Chia-Pan
2017-01-01
Reservoirs are important for households and impact the national economy. This paper proposed a time-series forecasting model based on estimating a missing value followed by variable selection to forecast the reservoir's water level. This study collected data from the Taiwan Shimen Reservoir as well as daily atmospheric data from 2008 to 2015. The two datasets are concatenated into an integrated dataset based on ordering of the data as a research dataset. The proposed time-series forecasting model summarily has three foci. First, this study uses five imputation methods to directly delete the missing value. Second, we identified the key variable via factor analysis and then deleted the unimportant variables sequentially via the variable selection method. Finally, the proposed model uses a Random Forest to build the forecasting model of the reservoir's water level. This was done to compare with the listing method under the forecasting error. These experimental results indicate that the Random Forest forecasting model when applied to variable selection with full variables has better forecasting performance than the listing model. In addition, this experiment shows that the proposed variable selection can help determine five forecast methods used here to improve the forecasting capability.
Future mission studies: Forecasting solar flux directly from its chaotic time series
Ashrafi, S.
1991-01-01
The mathematical structure of the programs written to construct a nonlinear predictive model to forecast solar flux directly from its time series without reference to any underlying solar physics is presented. This method and the programs are written so that one could apply the same technique to forecast other chaotic time series, such as geomagnetic data, attitude and orbit data, and even financial indexes and stock market data. Perhaps the most important application of this technique to flight dynamics is to model Goddard Trajectory Determination System (GTDS) output of residues between observed position of spacecraft and calculated position with no drag (drag flag = off). This would result in a new model of drag working directly from observed data.
A multivariate time series approach to modeling and forecasting demand in the emergency department.
Jones, Spencer S; Evans, R Scott; Allen, Todd L; Thomas, Alun; Haug, Peter J; Welch, Shari J; Snow, Gregory L
2009-02-01
The goals of this investigation were to study the temporal relationships between the demands for key resources in the emergency department (ED) and the inpatient hospital, and to develop multivariate forecasting models. Hourly data were collected from three diverse hospitals for the year 2006. Descriptive analysis and model fitting were carried out using graphical and multivariate time series methods. Multivariate models were compared to a univariate benchmark model in terms of their ability to provide out-of-sample forecasts of ED census and the demands for diagnostic resources. Descriptive analyses revealed little temporal interaction between the demand for inpatient resources and the demand for ED resources at the facilities considered. Multivariate models provided more accurate forecasts of ED census and of the demands for diagnostic resources. Our results suggest that multivariate time series models can be used to reliably forecast ED patient census; however, forecasts of the demands for diagnostic resources were not sufficiently reliable to be useful in the clinical setting.
A Time-Series Water Level Forecasting Model Based on Imputation and Variable Selection Method
Directory of Open Access Journals (Sweden)
Jun-He Yang
2017-01-01
Full Text Available Reservoirs are important for households and impact the national economy. This paper proposed a time-series forecasting model based on estimating a missing value followed by variable selection to forecast the reservoir’s water level. This study collected data from the Taiwan Shimen Reservoir as well as daily atmospheric data from 2008 to 2015. The two datasets are concatenated into an integrated dataset based on ordering of the data as a research dataset. The proposed time-series forecasting model summarily has three foci. First, this study uses five imputation methods to directly delete the missing value. Second, we identified the key variable via factor analysis and then deleted the unimportant variables sequentially via the variable selection method. Finally, the proposed model uses a Random Forest to build the forecasting model of the reservoir’s water level. This was done to compare with the listing method under the forecasting error. These experimental results indicate that the Random Forest forecasting model when applied to variable selection with full variables has better forecasting performance than the listing model. In addition, this experiment shows that the proposed variable selection can help determine five forecast methods used here to improve the forecasting capability.
Wang, Jun; Zhou, Bi-hua; Zhou, Shu-dao; Sheng, Zheng
2015-01-01
The paper proposes a novel function expression method to forecast chaotic time series, using an improved genetic-simulated annealing (IGSA) algorithm to establish the optimum function expression that describes the behavior of time series. In order to deal with the weakness associated with the genetic algorithm, the proposed algorithm incorporates the simulated annealing operation which has the strong local search ability into the genetic algorithm to enhance the performance of optimization; besides, the fitness function and genetic operators are also improved. Finally, the method is applied to the chaotic time series of Quadratic and Rossler maps for validation. The effect of noise in the chaotic time series is also studied numerically. The numerical results verify that the method can forecast chaotic time series with high precision and effectiveness, and the forecasting precision with certain noise is also satisfactory. It can be concluded that the IGSA algorithm is energy-efficient and superior.
Forecasting Long Memory Series Subject to Structural Change
DEFF Research Database (Denmark)
Dias, Gustavo Fruet; Papailias, Fotis
A two-stage forecasting approach for long memory time series is introduced. In the first step we estimate the fractional exponent and, applying the fractional differencing operator, we obtain the underlying weakly dependent series. In the second step, we perform the multi-step ahead forecasts...... for the weakly dependent series and obtain their long memory counterparts by applying the fractional cumulation operator. The methodology applies to stationary and nonstationary cases. Simulations and an application to seven time series provide evidence that the new methodology is more robust to structural...... change and yields good forecasting results....
Multiple Time Series Forecasting Using Quasi-Randomized Functional Link Neural Networks
Directory of Open Access Journals (Sweden)
Thierry Moudiki
2018-03-01
Full Text Available We are interested in obtaining forecasts for multiple time series, by taking into account the potential nonlinear relationships between their observations. For this purpose, we use a specific type of regression model on an augmented dataset of lagged time series. Our model is inspired by dynamic regression models (Pankratz 2012, with the response variable’s lags included as predictors, and is known as Random Vector Functional Link (RVFL neural networks. The RVFL neural networks have been successfully applied in the past, to solving regression and classification problems. The novelty of our approach is to apply an RVFL model to multivariate time series, under two separate regularization constraints on the regression parameters.
Nonlinear time series modeling and forecasting the seismic data of the Hindu Kush region
Khan, Muhammad Yousaf; Mittnik, Stefan
2018-01-01
In this study, we extended the application of linear and nonlinear time models in the field of earthquake seismology and examined the out-of-sample forecast accuracy of linear Autoregressive (AR), Autoregressive Conditional Duration (ACD), Self-Exciting Threshold Autoregressive (SETAR), Threshold Autoregressive (TAR), Logistic Smooth Transition Autoregressive (LSTAR), Additive Autoregressive (AAR), and Artificial Neural Network (ANN) models for seismic data of the Hindu Kush region. We also extended the previous studies by using Vector Autoregressive (VAR) and Threshold Vector Autoregressive (TVAR) models and compared their forecasting accuracy with linear AR model. Unlike previous studies that typically consider the threshold model specifications by using internal threshold variable, we specified these models with external transition variables and compared their out-of-sample forecasting performance with the linear benchmark AR model. The modeling results show that time series models used in the present study are capable of capturing the dynamic structure present in the seismic data. The point forecast results indicate that the AR model generally outperforms the nonlinear models. However, in some cases, threshold models with external threshold variables specification produce more accurate forecasts, indicating that specification of threshold time series models is of crucial importance. For raw seismic data, the ACD model does not show an improved out-of-sample forecasting performance over the linear AR model. The results indicate that the AR model is the best forecasting device to model and forecast the raw seismic data of the Hindu Kush region.
Directory of Open Access Journals (Sweden)
Yolcu Ufuk
2016-06-01
Full Text Available Fuzzy time series methods based on the fuzzy set theory proposed by Zadeh (1965 was first introduced by Song and Chissom (1993. Since fuzzy time series methods do not have the assumptions that traditional time series do and have effective forecasting performance, the interest on fuzzy time series approaches is increasing rapidly. Fuzzy time series methods have been used in almost all areas, such as environmental science, economy and finance. The concepts of labour force participation and unemployment have great importance in terms of both the economy and sociology of countries. For this reason there are many studies on their forecasting. In this study, we aim to forecast the labour force participation and unemployment rate in Poland and Turkey using different fuzzy time series methods.
Toward automatic time-series forecasting using neural networks.
Yan, Weizhong
2012-07-01
Over the past few decades, application of artificial neural networks (ANN) to time-series forecasting (TSF) has been growing rapidly due to several unique features of ANN models. However, to date, a consistent ANN performance over different studies has not been achieved. Many factors contribute to the inconsistency in the performance of neural network models. One such factor is that ANN modeling involves determining a large number of design parameters, and the current design practice is essentially heuristic and ad hoc, this does not exploit the full potential of neural networks. Systematic ANN modeling processes and strategies for TSF are, therefore, greatly needed. Motivated by this need, this paper attempts to develop an automatic ANN modeling scheme. It is based on the generalized regression neural network (GRNN), a special type of neural network. By taking advantage of several GRNN properties (i.e., a single design parameter and fast learning) and by incorporating several design strategies (e.g., fusing multiple GRNNs), we have been able to make the proposed modeling scheme to be effective for modeling large-scale business time series. The initial model was entered into the NN3 time-series competition. It was awarded the best prediction on the reduced dataset among approximately 60 different models submitted by scholars worldwide.
Time Series Forecasting of the Number of Malaysia Airlines and AirAsia Passengers
Asrah, N. M.; Nor, M. E.; Rahim, S. N. A.; Leng, W. K.
2018-04-01
The standard practice in forecasting process involved by fitting a model and further analysis on the residuals. If we know the distributional behaviour of the time series data, it can help us to directly analyse the model identification, parameter estimation, and model checking. In this paper, we want to compare the distributional behaviour data from the number of Malaysia Airlines (MAS) and AirAsia passenger’s. From the previous research, the AirAsia passengers are govern by geometric Brownian motion (GBM). The data were normally distributed, stationary and independent. Then, GBM was used to forecast the number of AirAsia passenger’s. The same methods were applied to MAS data and the results then were compared. Unfortunately, the MAS data were not govern by GBM. Then, the standard approach in time series forecasting will be applied to MAS data. From this comparison, we can conclude that the number of AirAsia passengers are always in peak season rather than MAS passengers.
Autoregressive Prediction with Rolling Mechanism for Time Series Forecasting with Small Sample Size
Directory of Open Access Journals (Sweden)
Zhihua Wang
2014-01-01
Full Text Available Reasonable prediction makes significant practical sense to stochastic and unstable time series analysis with small or limited sample size. Motivated by the rolling idea in grey theory and the practical relevance of very short-term forecasting or 1-step-ahead prediction, a novel autoregressive (AR prediction approach with rolling mechanism is proposed. In the modeling procedure, a new developed AR equation, which can be used to model nonstationary time series, is constructed in each prediction step. Meanwhile, the data window, for the next step ahead forecasting, rolls on by adding the most recent derived prediction result while deleting the first value of the former used sample data set. This rolling mechanism is an efficient technique for its advantages of improved forecasting accuracy, applicability in the case of limited and unstable data situations, and requirement of little computational effort. The general performance, influence of sample size, nonlinearity dynamic mechanism, and significance of the observed trends, as well as innovation variance, are illustrated and verified with Monte Carlo simulations. The proposed methodology is then applied to several practical data sets, including multiple building settlement sequences and two economic series.
Klevtsov, S. I.
2018-05-01
The impact of physical factors, such as temperature and others, leads to a change in the parameters of the technical object. Monitoring the change of parameters is necessary to prevent a dangerous situation. The control is carried out in real time. To predict the change in the parameter, a time series is used in this paper. Forecasting allows one to determine the possibility of a dangerous change in a parameter before the moment when this change occurs. The control system in this case has more time to prevent a dangerous situation. A simple time series was chosen. In this case, the algorithm is simple. The algorithm is executed in the microprocessor module in the background. The efficiency of using the time series is affected by its characteristics, which must be adjusted. In the work, the influence of these characteristics on the error of prediction of the controlled parameter was studied. This takes into account the behavior of the parameter. The values of the forecast lag are determined. The results of the research, in the case of their use, will improve the efficiency of monitoring the technical object during its operation.
International Nuclear Information System (INIS)
Lefieux, V.
2007-10-01
Reseau de Transport d'Electricite (RTE), in charge of operating the French electric transportation grid, needs an accurate forecast of the power consumption in order to operate it correctly. The forecasts used everyday result from a model combining a nonlinear parametric regression and a SARIMA model. In order to obtain an adaptive forecasting model, nonparametric forecasting methods have already been tested without real success. In particular, it is known that a nonparametric predictor behaves badly with a great number of explanatory variables, what is commonly called the curse of dimensionality. Recently, semi parametric methods which improve the pure nonparametric approach have been proposed to estimate a regression function. Based on the concept of 'dimension reduction', one those methods (called MAVE : Moving Average -conditional- Variance Estimate) can apply to time series. We study empirically its effectiveness to predict the future values of an autoregressive time series. We then adapt this method, from a practical point of view, to forecast power consumption. We propose a partially linear semi parametric model, based on the MAVE method, which allows to take into account simultaneously the autoregressive aspect of the problem and the exogenous variables. The proposed estimation procedure is practically efficient. (author)
Linden, Ariel
2018-05-11
Interrupted time series analysis (ITSA) is an evaluation methodology in which a single treatment unit's outcome is studied serially over time and the intervention is expected to "interrupt" the level and/or trend of that outcome. ITSA is commonly evaluated using methods which may produce biased results if model assumptions are violated. In this paper, treatment effects are alternatively assessed by using forecasting methods to closely fit the preintervention observations and then forecast the post-intervention trend. A treatment effect may be inferred if the actual post-intervention observations diverge from the forecasts by some specified amount. The forecasting approach is demonstrated using the effect of California's Proposition 99 for reducing cigarette sales. Three forecast models are fit to the preintervention series-linear regression (REG), Holt-Winters (HW) non-seasonal smoothing, and autoregressive moving average (ARIMA)-and forecasts are generated into the post-intervention period. The actual observations are then compared with the forecasts to assess intervention effects. The preintervention data were fit best by HW, followed closely by ARIMA. REG fit the data poorly. The actual post-intervention observations were above the forecasts in HW and ARIMA, suggesting no intervention effect, but below the forecasts in the REG (suggesting a treatment effect), thereby raising doubts about any definitive conclusion of a treatment effect. In a single-group ITSA, treatment effects are likely to be biased if the model is misspecified. Therefore, evaluators should consider using forecast models to accurately fit the preintervention data and generate plausible counterfactual forecasts, thereby improving causal inference of treatment effects in single-group ITSA studies. © 2018 John Wiley & Sons, Ltd.
Modeling the impact of forecast-based regime switches on macroeconomic time series
K. Bel (Koen); R. Paap (Richard)
2013-01-01
textabstractForecasts of key macroeconomic variables may lead to policy changes of governments, central banks and other economic agents. Policy changes in turn lead to structural changes in macroeconomic time series models. To describe this phenomenon we introduce a logistic smooth transition
High-order fuzzy time-series based on multi-period adaptation model for forecasting stock markets
Chen, Tai-Liang; Cheng, Ching-Hsue; Teoh, Hia-Jong
2008-02-01
Stock investors usually make their short-term investment decisions according to recent stock information such as the late market news, technical analysis reports, and price fluctuations. To reflect these short-term factors which impact stock price, this paper proposes a comprehensive fuzzy time-series, which factors linear relationships between recent periods of stock prices and fuzzy logical relationships (nonlinear relationships) mined from time-series into forecasting processes. In empirical analysis, the TAIEX (Taiwan Stock Exchange Capitalization Weighted Stock Index) and HSI (Heng Seng Index) are employed as experimental datasets, and four recent fuzzy time-series models, Chen’s (1996), Yu’s (2005), Cheng’s (2006) and Chen’s (2007), are used as comparison models. Besides, to compare with conventional statistic method, the method of least squares is utilized to estimate the auto-regressive models of the testing periods within the databases. From analysis results, the performance comparisons indicate that the multi-period adaptation model, proposed in this paper, can effectively improve the forecasting performance of conventional fuzzy time-series models which only factor fuzzy logical relationships in forecasting processes. From the empirical study, the traditional statistic method and the proposed model both reveal that stock price patterns in the Taiwan stock and Hong Kong stock markets are short-term.
International Work-Conference on Time Series
Pomares, Héctor; Valenzuela, Olga
2017-01-01
This volume of selected and peer-reviewed contributions on the latest developments in time series analysis and forecasting updates the reader on topics such as analysis of irregularly sampled time series, multi-scale analysis of univariate and multivariate time series, linear and non-linear time series models, advanced time series forecasting methods, applications in time series analysis and forecasting, advanced methods and online learning in time series and high-dimensional and complex/big data time series. The contributions were originally presented at the International Work-Conference on Time Series, ITISE 2016, held in Granada, Spain, June 27-29, 2016. The series of ITISE conferences provides a forum for scientists, engineers, educators and students to discuss the latest ideas and implementations in the foundations, theory, models and applications in the field of time series analysis and forecasting. It focuses on interdisciplinary and multidisciplinary rese arch encompassing the disciplines of comput...
Zhang, Hong; Zhang, Sheng; Wang, Ping; Qin, Yuzhe; Wang, Huifeng
2017-07-01
Particulate matter with aerodynamic diameter below 10 μm (PM 10 ) forecasting is difficult because of the uncertainties in describing the emission and meteorological fields. This paper proposed a wavelet-ARMA/ARIMA model to forecast the short-term series of the PM 10 concentrations. It was evaluated by experiments using a 10-year data set of daily PM 10 concentrations from 4 stations located in Taiyuan, China. The results indicated the following: (1) PM 10 concentrations of Taiyuan had a decreasing trend during 2005 to 2012 but increased in 2013. PM 10 concentrations had an obvious seasonal fluctuation related to coal-fired heating in winter and early spring. (2) Spatial differences among the four stations showed that the PM 10 concentrations in industrial and heavily trafficked areas were higher than those in residential and suburb areas. (3) Wavelet analysis revealed that the trend variation and the changes of the PM 10 concentration of Taiyuan were complicated. (4) The proposed wavelet-ARIMA model could be efficiently and successfully applied to the PM 10 forecasting field. Compared with the traditional ARMA/ARIMA methods, this wavelet-ARMA/ARIMA method could effectively reduce the forecasting error, improve the prediction accuracy, and realize multiple-time-scale prediction. Wavelet analysis can filter noisy signals and identify the variation trend and the fluctuation of the PM 10 time-series data. Wavelet decomposition and reconstruction reduce the nonstationarity of the PM 10 time-series data, and thus improve the accuracy of the prediction. This paper proposed a wavelet-ARMA/ARIMA model to forecast the PM 10 time series. Compared with the traditional ARMA/ARIMA method, this wavelet-ARMA/ARIMA method could effectively reduce the forecasting error, improve the prediction accuracy, and realize multiple-time-scale prediction. The proposed model could be efficiently and successfully applied to the PM 10 forecasting field.
Forecasting incidence of dengue in Rajasthan, using time series analyses.
Bhatnagar, Sunil; Lal, Vivek; Gupta, Shiv D; Gupta, Om P
2012-01-01
To develop a prediction model for dengue fever/dengue haemorrhagic fever (DF/DHF) using time series data over the past decade in Rajasthan and to forecast monthly DF/DHF incidence for 2011. Seasonal autoregressive integrated moving average (SARIMA) model was used for statistical modeling. During January 2001 to December 2010, the reported DF/DHF cases showed a cyclical pattern with seasonal variation. SARIMA (0,0,1) (0,1,1) 12 model had the lowest normalized Bayesian information criteria (BIC) of 9.426 and mean absolute percentage error (MAPE) of 263.361 and appeared to be the best model. The proportion of variance explained by the model was 54.3%. Adequacy of the model was established through Ljung-Box test (Q statistic 4.910 and P-value 0.996), which showed no significant correlation between residuals at different lag times. The forecast for the year 2011 showed a seasonal peak in the month of October with an estimated 546 cases. Application of SARIMA model may be useful for forecast of cases and impending outbreaks of DF/DHF and other infectious diseases, which exhibit seasonal pattern.
Road safety forecasts in five European countries using structural time series models.
Antoniou, Constantinos; Papadimitriou, Eleonora; Yannis, George
2014-01-01
Modeling road safety development is a complex task and needs to consider both the quantifiable impact of specific parameters as well as the underlying trends that cannot always be measured or observed. The objective of this research is to apply structural time series models for obtaining reliable medium- to long-term forecasts of road traffic fatality risk using data from 5 countries with different characteristics from all over Europe (Cyprus, Greece, Hungary, Norway, and Switzerland). Two structural time series models are considered: (1) the local linear trend model and the (2) latent risk time series model. Furthermore, a structured decision tree for the selection of the applicable model for each situation (developed within the Road Safety Data, Collection, Transfer and Analysis [DaCoTA] research project, cofunded by the European Commission) is outlined. First, the fatality and exposure data that are used for the development of the models are presented and explored. Then, the modeling process is presented, including the model selection process, introduction of intervention variables, and development of mobility scenarios. The forecasts using the developed models appear to be realistic and within acceptable confidence intervals. The proposed methodology is proved to be very efficient for handling different cases of data availability and quality, providing an appropriate alternative from the family of structural time series models in each country. A concluding section providing perspectives and directions for future research is presented.
Multi-Step Time Series Forecasting with an Ensemble of Varied Length Mixture Models.
Ouyang, Yicun; Yin, Hujun
2018-05-01
Many real-world problems require modeling and forecasting of time series, such as weather temperature, electricity demand, stock prices and foreign exchange (FX) rates. Often, the tasks involve predicting over a long-term period, e.g. several weeks or months. Most existing time series models are inheritably for one-step prediction, that is, predicting one time point ahead. Multi-step or long-term prediction is difficult and challenging due to the lack of information and uncertainty or error accumulation. The main existing approaches, iterative and independent, either use one-step model recursively or treat the multi-step task as an independent model. They generally perform poorly in practical applications. In this paper, as an extension of the self-organizing mixture autoregressive (AR) model, the varied length mixture (VLM) models are proposed to model and forecast time series over multi-steps. The key idea is to preserve the dependencies between the time points within the prediction horizon. Training data are segmented to various lengths corresponding to various forecasting horizons, and the VLM models are trained in a self-organizing fashion on these segments to capture these dependencies in its component AR models of various predicting horizons. The VLM models form a probabilistic mixture of these varied length models. A combination of short and long VLM models and an ensemble of them are proposed to further enhance the prediction performance. The effectiveness of the proposed methods and their marked improvements over the existing methods are demonstrated through a number of experiments on synthetic data, real-world FX rates and weather temperatures.
Mohammed, Emad A; Naugler, Christopher
2017-01-01
Demand forecasting is the area of predictive analytics devoted to predicting future volumes of services or consumables. Fair understanding and estimation of how demand will vary facilitates the optimal utilization of resources. In a medical laboratory, accurate forecasting of future demand, that is, test volumes, can increase efficiency and facilitate long-term laboratory planning. Importantly, in an era of utilization management initiatives, accurately predicted volumes compared to the realized test volumes can form a precise way to evaluate utilization management initiatives. Laboratory test volumes are often highly amenable to forecasting by time-series models; however, the statistical software needed to do this is generally either expensive or highly technical. In this paper, we describe an open-source web-based software tool for time-series forecasting and explain how to use it as a demand forecasting tool in clinical laboratories to estimate test volumes. This tool has three different models, that is, Holt-Winters multiplicative, Holt-Winters additive, and simple linear regression. Moreover, these models are ranked and the best one is highlighted. This tool will allow anyone with historic test volume data to model future demand.
Directory of Open Access Journals (Sweden)
Emad A Mohammed
2017-01-01
Full Text Available Background: Demand forecasting is the area of predictive analytics devoted to predicting future volumes of services or consumables. Fair understanding and estimation of how demand will vary facilitates the optimal utilization of resources. In a medical laboratory, accurate forecasting of future demand, that is, test volumes, can increase efficiency and facilitate long-term laboratory planning. Importantly, in an era of utilization management initiatives, accurately predicted volumes compared to the realized test volumes can form a precise way to evaluate utilization management initiatives. Laboratory test volumes are often highly amenable to forecasting by time-series models; however, the statistical software needed to do this is generally either expensive or highly technical. Method: In this paper, we describe an open-source web-based software tool for time-series forecasting and explain how to use it as a demand forecasting tool in clinical laboratories to estimate test volumes. Results: This tool has three different models, that is, Holt-Winters multiplicative, Holt-Winters additive, and simple linear regression. Moreover, these models are ranked and the best one is highlighted. Conclusion: This tool will allow anyone with historic test volume data to model future demand.
Time series analysis for psychological research: examining and forecasting change.
Jebb, Andrew T; Tay, Louis; Wang, Wei; Huang, Qiming
2015-01-01
Psychological research has increasingly recognized the importance of integrating temporal dynamics into its theories, and innovations in longitudinal designs and analyses have allowed such theories to be formalized and tested. However, psychological researchers may be relatively unequipped to analyze such data, given its many characteristics and the general complexities involved in longitudinal modeling. The current paper introduces time series analysis to psychological research, an analytic domain that has been essential for understanding and predicting the behavior of variables across many diverse fields. First, the characteristics of time series data are discussed. Second, different time series modeling techniques are surveyed that can address various topics of interest to psychological researchers, including describing the pattern of change in a variable, modeling seasonal effects, assessing the immediate and long-term impact of a salient event, and forecasting future values. To illustrate these methods, an illustrative example based on online job search behavior is used throughout the paper, and a software tutorial in R for these analyses is provided in the Supplementary Materials.
Time series analysis for psychological research: examining and forecasting change
Jebb, Andrew T.; Tay, Louis; Wang, Wei; Huang, Qiming
2015-01-01
Psychological research has increasingly recognized the importance of integrating temporal dynamics into its theories, and innovations in longitudinal designs and analyses have allowed such theories to be formalized and tested. However, psychological researchers may be relatively unequipped to analyze such data, given its many characteristics and the general complexities involved in longitudinal modeling. The current paper introduces time series analysis to psychological research, an analytic domain that has been essential for understanding and predicting the behavior of variables across many diverse fields. First, the characteristics of time series data are discussed. Second, different time series modeling techniques are surveyed that can address various topics of interest to psychological researchers, including describing the pattern of change in a variable, modeling seasonal effects, assessing the immediate and long-term impact of a salient event, and forecasting future values. To illustrate these methods, an illustrative example based on online job search behavior is used throughout the paper, and a software tutorial in R for these analyses is provided in the Supplementary Materials. PMID:26106341
Multidimensional k-nearest neighbor model based on EEMD for financial time series forecasting
Zhang, Ningning; Lin, Aijing; Shang, Pengjian
2017-07-01
In this paper, we propose a new two-stage methodology that combines the ensemble empirical mode decomposition (EEMD) with multidimensional k-nearest neighbor model (MKNN) in order to forecast the closing price and high price of the stocks simultaneously. The modified algorithm of k-nearest neighbors (KNN) has an increasingly wide application in the prediction of all fields. Empirical mode decomposition (EMD) decomposes a nonlinear and non-stationary signal into a series of intrinsic mode functions (IMFs), however, it cannot reveal characteristic information of the signal with much accuracy as a result of mode mixing. So ensemble empirical mode decomposition (EEMD), an improved method of EMD, is presented to resolve the weaknesses of EMD by adding white noise to the original data. With EEMD, the components with true physical meaning can be extracted from the time series. Utilizing the advantage of EEMD and MKNN, the new proposed ensemble empirical mode decomposition combined with multidimensional k-nearest neighbor model (EEMD-MKNN) has high predictive precision for short-term forecasting. Moreover, we extend this methodology to the case of two-dimensions to forecast the closing price and high price of the four stocks (NAS, S&P500, DJI and STI stock indices) at the same time. The results indicate that the proposed EEMD-MKNN model has a higher forecast precision than EMD-KNN, KNN method and ARIMA.
Directory of Open Access Journals (Sweden)
Jan Dempewolf
2014-10-01
Full Text Available Policy makers, government planners and agricultural market participants in Pakistan require accurate and timely information about wheat yield and production. Punjab Province is by far the most important wheat producing region in the country. The manual collection of field data and data processing for crop forecasting by the provincial government requires significant amounts of time before official reports can be released. Several studies have shown that wheat yield can be effectively forecast using satellite remote sensing data. In this study, we developed a methodology for estimating wheat yield and area for Punjab Province from freely available Landsat and MODIS satellite imagery approximately six weeks before harvest. Wheat yield was derived by regressing reported yield values against time series of four different peak-season MODIS-derived vegetation indices. We also tested deriving wheat area from the same MODIS time series using a regression-tree approach. Among the four evaluated indices, WDRVI provided more consistent and accurate yield forecasts compared to NDVI, EVI2 and saturation-adjusted normalized difference vegetation index (SANDVI. The lowest RMSE values at the district level for forecast versus reported yield were found when using six or more years of training data. Forecast yield for the 2007/2008 to 2012/2013 growing seasons were within 0.2% and 11.5% of final reported values. Absolute deviations of wheat area and production forecasts from reported values were slightly greater compared to using the previous year's or the three- or six-year moving average values, implying that 250-m MODIS data does not provide sufficient spatial resolution for providing improved wheat area and production forecasts.
Curceac, S.; Ternynck, C.; Ouarda, T.
2015-12-01
Over the past decades, a substantial amount of research has been conducted to model and forecast climatic variables. In this study, Nonparametric Functional Data Analysis (NPFDA) methods are applied to forecast air temperature and wind speed time series in Abu Dhabi, UAE. The dataset consists of hourly measurements recorded for a period of 29 years, 1982-2010. The novelty of the Functional Data Analysis approach is in expressing the data as curves. In the present work, the focus is on daily forecasting and the functional observations (curves) express the daily measurements of the above mentioned variables. We apply a non-linear regression model with a functional non-parametric kernel estimator. The computation of the estimator is performed using an asymmetrical quadratic kernel function for local weighting based on the bandwidth obtained by a cross validation procedure. The proximities between functional objects are calculated by families of semi-metrics based on derivatives and Functional Principal Component Analysis (FPCA). Additionally, functional conditional mode and functional conditional median estimators are applied and the advantages of combining their results are analysed. A different approach employs a SARIMA model selected according to the minimum Akaike (AIC) and Bayessian (BIC) Information Criteria and based on the residuals of the model. The performance of the models is assessed by calculating error indices such as the root mean square error (RMSE), relative RMSE, BIAS and relative BIAS. The results indicate that the NPFDA models provide more accurate forecasts than the SARIMA models. Key words: Nonparametric functional data analysis, SARIMA, time series forecast, air temperature, wind speed
Forecasting Container Throughput at the Doraleh Port in Djibouti through Time Series Analysis
Mohamed Ismael, Hawa; Vandyck, George Kobina
The Doraleh Container Terminal (DCT) located in Djibouti has been noted as the most technologically advanced container terminal on the African continent. DCT's strategic location at the crossroads of the main shipping lanes connecting Asia, Africa and Europe put it in a unique position to provide important shipping services to vessels plying that route. This paper aims to forecast container throughput through the Doraleh Container Port in Djibouti by Time Series Analysis. A selection of univariate forecasting models has been used, namely Triple Exponential Smoothing Model, Grey Model and Linear Regression Model. By utilizing the above three models and their combination, the forecast of container throughput through the Doraleh port was realized. A comparison of the different forecasting results of the three models, in addition to the combination forecast is then undertaken, based on commonly used evaluation criteria Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE). The study found that the Linear Regression forecasting Model was the best prediction method for forecasting the container throughput, since its forecast error was the least. Based on the regression model, a ten (10) year forecast for container throughput at DCT has been made.
A comparison of various forecasting techniques applied to mean hourly wind speed time series
Energy Technology Data Exchange (ETDEWEB)
Sfetsos, A. [7 Pirsou Street, Athens (Greece)
2000-09-01
This paper presents a comparison of various forecasting approaches, using time series analysis, on mean hourly wind speed data. In addition to the traditional linear (ARMA) models and the commonly used feed forward and recurrent neural networks, other approaches are also examined including the Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Neural Logic Networks. The developed models are evaluated for their ability to produce accurate and fast forecasts. (Author)
Stochastic Simulation and Forecast of Hydrologic Time Series Based on Probabilistic Chaos Expansion
Li, Z.; Ghaith, M.
2017-12-01
Hydrological processes are characterized by many complex features, such as nonlinearity, dynamics and uncertainty. How to quantify and address such complexities and uncertainties has been a challenging task for water engineers and managers for decades. To support robust uncertainty analysis, an innovative approach for the stochastic simulation and forecast of hydrologic time series is developed is this study. Probabilistic Chaos Expansions (PCEs) are established through probabilistic collocation to tackle uncertainties associated with the parameters of traditional hydrological models. The uncertainties are quantified in model outputs as Hermite polynomials with regard to standard normal random variables. Sequentially, multivariate analysis techniques are used to analyze the complex nonlinear relationships between meteorological inputs (e.g., temperature, precipitation, evapotranspiration, etc.) and the coefficients of the Hermite polynomials. With the established relationships between model inputs and PCE coefficients, forecasts of hydrologic time series can be generated and the uncertainties in the future time series can be further tackled. The proposed approach is demonstrated using a case study in China and is compared to a traditional stochastic simulation technique, the Markov-Chain Monte-Carlo (MCMC) method. Results show that the proposed approach can serve as a reliable proxy to complicated hydrological models. It can provide probabilistic forecasting in a more computationally efficient manner, compared to the traditional MCMC method. This work provides technical support for addressing uncertainties associated with hydrological modeling and for enhancing the reliability of hydrological modeling results. Applications of the developed approach can be extended to many other complicated geophysical and environmental modeling systems to support the associated uncertainty quantification and risk analysis.
Karpušenkaitė, Aistė; Ruzgas, Tomas; Denafas, Gintaras
2018-05-01
The aim of the study was to create a hybrid forecasting method that could produce higher accuracy forecasts than previously used 'pure' time series methods. Mentioned methods were already tested with total automotive waste, hazardous automotive waste, and total medical waste generation, but demonstrated at least a 6% error rate in different cases and efforts were made to decrease it even more. Newly developed hybrid models used a random start generation method to incorporate different time-series advantages and it helped to increase the accuracy of forecasts by 3%-4% in hazardous automotive waste and total medical waste generation cases; the new model did not increase the accuracy of total automotive waste generation forecasts. Developed models' abilities to forecast short- and mid-term forecasts were tested using prediction horizon.
International Work-Conference on Time Series
Pomares, Héctor
2016-01-01
This volume presents selected peer-reviewed contributions from The International Work-Conference on Time Series, ITISE 2015, held in Granada, Spain, July 1-3, 2015. It discusses topics in time series analysis and forecasting, advanced methods and online learning in time series, high-dimensional and complex/big data time series as well as forecasting in real problems. The International Work-Conferences on Time Series (ITISE) provide a forum for scientists, engineers, educators and students to discuss the latest ideas and implementations in the foundations, theory, models and applications in the field of time series analysis and forecasting. It focuses on interdisciplinary and multidisciplinary research encompassing the disciplines of computer science, mathematics, statistics and econometrics.
Forecasting malaria cases using climatic factors in delhi, India: a time series analysis.
Kumar, Varun; Mangal, Abha; Panesar, Sanjeet; Yadav, Geeta; Talwar, Richa; Raut, Deepak; Singh, Saudan
2014-01-01
Background. Malaria still remains a public health problem in developing countries and changing environmental and climatic factors pose the biggest challenge in fighting against the scourge of malaria. Therefore, the study was designed to forecast malaria cases using climatic factors as predictors in Delhi, India. Methods. The total number of monthly cases of malaria slide positives occurring from January 2006 to December 2013 was taken from the register maintained at the malaria clinic at Rural Health Training Centre (RHTC), Najafgarh, Delhi. Climatic data of monthly mean rainfall, relative humidity, and mean maximum temperature were taken from Regional Meteorological Centre, Delhi. Expert modeler of SPSS ver. 21 was used for analyzing the time series data. Results. Autoregressive integrated moving average, ARIMA (0,1,1) (0,1,0)(12), was the best fit model and it could explain 72.5% variability in the time series data. Rainfall (P value = 0.004) and relative humidity (P value = 0.001) were found to be significant predictors for malaria transmission in the study area. Seasonal adjusted factor (SAF) for malaria cases shows peak during the months of August and September. Conclusion. ARIMA models of time series analysis is a simple and reliable tool for producing reliable forecasts for malaria in Delhi, India.
On probabilistic forecasting of wind power time-series
DEFF Research Database (Denmark)
Pinson, Pierre
power dynamics. In both cases, the model parameters are adaptively and recursively estimated, time-adaptativity being the result of exponential forgetting of past observations. The probabilistic forecasting methodology is applied at the Horns Rev wind farm in Denmark, for 10-minute ahead probabilistic...... forecasting of wind power generation. Probabilistic forecasts generated from the proposed methodology clearly have higher skill than those obtained from a classical Gaussian assumption about wind power predictive densities. Corresponding point forecasts also exhibit significantly lower error criteria....
Directory of Open Access Journals (Sweden)
David Afolabi
2017-11-01
Full Text Available The importance of an interference-less machine learning scheme in time series prediction is crucial, as an oversight can have a negative cumulative effect, especially when predicting many steps ahead of the currently available data. The on-going research on noise elimination in time series forecasting has led to a successful approach of decomposing the data sequence into component trends to identify noise-inducing information. The empirical mode decomposition method separates the time series/signal into a set of intrinsic mode functions ranging from high to low frequencies, which can be summed up to reconstruct the original data. The usual assumption that random noises are only contained in the high-frequency component has been shown not to be the case, as observed in our previous findings. The results from that experiment reveal that noise can be present in a low frequency component, and this motivates the newly-proposed algorithm. Additionally, to prevent the erosion of periodic trends and patterns within the series, we perform the learning of local and global trends separately in a hierarchical manner which succeeds in detecting and eliminating short/long term noise. The algorithm is tested on four datasets from financial market data and physical science data. The simulation results are compared with the conventional and state-of-the-art approaches for time series machine learning, such as the non-linear autoregressive neural network and the long short-term memory recurrent neural network, respectively. Statistically significant performance gains are recorded when the meta-learning algorithm for noise reduction is used in combination with these artificial neural networks. For time series data which cannot be decomposed into meaningful trends, applying the moving average method to create meta-information for guiding the learning process is still better than the traditional approach. Therefore, this new approach is applicable to the forecasting
Electric power demand forecasting using interval time series. A comparison between VAR and iMLP
International Nuclear Information System (INIS)
Garcia-Ascanio, Carolina; Mate, Carlos
2010-01-01
Electric power demand forecasts play an essential role in the electric industry, as they provide the basis for making decisions in power system planning and operation. A great variety of mathematical methods have been used for demand forecasting. The development and improvement of appropriate mathematical tools will lead to more accurate demand forecasting techniques. In order to forecast the monthly electric power demand per hour in Spain for 2 years, this paper presents a comparison between a new forecasting approach considering vector autoregressive (VAR) forecasting models applied to interval time series (ITS) and the iMLP, the multi-layer perceptron model adapted to interval data. In the proposed comparison, for the VAR approach two models are fitted per every hour, one composed of the centre (mid-point) and radius (half-range), and another one of the lower and upper bounds according to the interval representation assumed by the ITS in the learning set. In the case of the iMLP, only the model composed of the centre and radius is fitted. The other interval representation composed of the lower and upper bounds is obtained from the linear combination of the two. This novel approach, obtaining two bivariate models each hour, makes possible to establish, for different periods in the day, which interval representation is more accurate. Furthermore, the comparison between two different techniques adapted to interval time series allows us to determine the efficiency of these models in forecasting electric power demand. It is important to note that the iMLP technique has been selected for the comparison, as it has shown its accuracy in forecasting daily electricity price intervals. This work shows the ITS forecasting methods as a potential tool that will lead to a reduction in risk when making power system planning and operational decisions. (author)
Nonlinear techniques for forecasting solar activity directly from its time series
Ashrafi, S.; Roszman, L.; Cooley, J.
1993-01-01
This paper presents numerical techniques for constructing nonlinear predictive models to forecast solar flux directly from its time series. This approach makes it possible to extract dynamical in variants of our system without reference to any underlying solar physics. We consider the dynamical evolution of solar activity in a reconstructed phase space that captures the attractor (strange), give a procedure for constructing a predictor of future solar activity, and discuss extraction of dynamical invariants such as Lyapunov exponents and attractor dimension.
WANG, D.; Wang, Y.; Zeng, X.
2017-12-01
Accurate, fast forecasting of hydro-meteorological time series is presently a major challenge in drought and flood mitigation. This paper proposes a hybrid approach, Wavelet De-noising (WD) and Rank-Set Pair Analysis (RSPA), that takes full advantage of a combination of the two approaches to improve forecasts of hydro-meteorological time series. WD allows decomposition and reconstruction of a time series by the wavelet transform, and hence separation of the noise from the original series. RSPA, a more reliable and efficient version of Set Pair Analysis, is integrated with WD to form the hybrid WD-RSPA approach. Two types of hydro-meteorological data sets with different characteristics and different levels of human influences at some representative stations are used to illustrate the WD-RSPA approach. The approach is also compared to three other generic methods: the conventional Auto Regressive Integrated Moving Average (ARIMA) method, Artificial Neural Networks (ANNs) (BP-error Back Propagation, MLP-Multilayer Perceptron and RBF-Radial Basis Function), and RSPA alone. Nine error metrics are used to evaluate the model performance. The results show that WD-RSPA is accurate, feasible, and effective. In particular, WD-RSPA is found to be the best among the various generic methods compared in this paper, even when the extreme events are included within a time series.
Spaeder, M C; Fackler, J C
2012-04-01
Respiratory syncytial virus (RSV) is the most common cause of documented viral respiratory infections, and the leading cause of hospitalization, in young children. We performed a retrospective time-series analysis of all patients aged Forecasting models of weekly RSV incidence for the local community, inpatient paediatric hospital and paediatric intensive-care unit (PICU) were created. Ninety-five percent confidence intervals calculated around our models' 2-week forecasts were accurate to ±9·3, ±7·5 and ±1·5 cases/week for the local community, inpatient hospital and PICU, respectively. Our results suggest that time-series models may be useful tools in forecasting the burden of RSV infection at the local and institutional levels, helping communities and institutions to optimize distribution of resources based on the changing burden and severity of illness in their respective communities.
Stock price forecasting based on time series analysis
Chi, Wan Le
2018-05-01
Using the historical stock price data to set up a sequence model to explain the intrinsic relationship of data, the future stock price can forecasted. The used models are auto-regressive model, moving-average model and autoregressive-movingaverage model. The original data sequence of unit root test was used to judge whether the original data sequence was stationary. The non-stationary original sequence as a first order difference needed further processing. Then the stability of the sequence difference was re-inspected. If it is still non-stationary, the second order differential processing of the sequence is carried out. Autocorrelation diagram and partial correlation diagram were used to evaluate the parameters of the identified ARMA model, including coefficients of the model and model order. Finally, the model was used to forecast the fitting of the shanghai composite index daily closing price with precision. Results showed that the non-stationary original data series was stationary after the second order difference. The forecast value of shanghai composite index daily closing price was closer to actual value, indicating that the ARMA model in the paper was a certain accuracy.
Modeling and Forecasting of Water Demand in Isfahan Using Underlying Trend Concept and Time Series
Directory of Open Access Journals (Sweden)
H. Sadeghi
2016-02-01
Full Text Available Introduction: Accurate water demand modeling for the city is very important for forecasting and policies adoption related to water resources management. Thus, for future requirements of water estimation, forecasting and modeling, it is important to utilize models with little errors. Water has a special place among the basic human needs, because it not hampers human life. The importance of the issue of water management in the extraction and consumption, it is necessary as a basic need. Municipal water applications is include a variety of water demand for domestic, public, industrial and commercial. Predicting the impact of urban water demand in better planning of water resources in arid and semiarid regions are faced with water restrictions. Materials and Methods: One of the most important factors affecting the changing technological advances in production and demand functions, we must pay special attention to the layout pattern. Technology development is concerned not only technically, but also other aspects such as personal, non-economic factors (population, geographical and social factors can be analyzed. Model examined in this study, a regression model is composed of a series of structural components over time allows changed invisible accidentally. Explanatory variables technology (both crystalline and amorphous in a model according to which the material is said to be better, but because of the lack of measured variables over time can not be entered in the template. Model examined in this study, a regression model is composed of a series of structural component invisible accidentally changed over time allows. In this study, structural time series (STSM and ARMA time series models have been used to model and estimate the water demand in Isfahan. Moreover, in order to find the efficient procedure, both models have been compared to each other. The desired data in this research include water consumption in Isfahan, water price and the monthly pay
Time series regression and ARIMAX for forecasting currency flow at Bank Indonesia in Sulawesi region
Suharsono, Agus; Suhartono, Masyitha, Aulia; Anuravega, Arum
2015-12-01
The purpose of the study is to forecast the outflow and inflow of currency at Indonesian Central Bank or Bank Indonesia (BI) in Sulawesi Region. The currency outflow and inflow data tend to have a trend pattern which is influenced by calendar variation effects. Therefore, this research focuses to apply some forecasting methods that could handle calendar variation effects, i.e. Time Series Regression (TSR) and ARIMAX models, and compare the forecast accuracy with ARIMA model. The best model is selected based on the lowest of Root Mean Squares Errors (RMSE) at out-sample dataset. The results show that ARIMA is the best model for forecasting the currency outflow and inflow at South Sulawesi. Whereas, the best model for forecasting the currency outflow at Central Sulawesi and Southeast Sulawesi, and for forecasting the currency inflow at South Sulawesi and North Sulawesi is TSR. Additionally, ARIMAX is the best model for forecasting the currency outflow at North Sulawesi. Hence, the results show that more complex models do not neccessary yield more accurate forecast than the simpler one.
Wang, Dong; Borthwick, Alistair G; He, Handan; Wang, Yuankun; Zhu, Jieyu; Lu, Yuan; Xu, Pengcheng; Zeng, Xiankui; Wu, Jichun; Wang, Lachun; Zou, Xinqing; Liu, Jiufu; Zou, Ying; He, Ruimin
2018-01-01
Accurate, fast forecasting of hydro-meteorological time series is presently a major challenge in drought and flood mitigation. This paper proposes a hybrid approach, wavelet de-noising (WD) and Rank-Set Pair Analysis (RSPA), that takes full advantage of a combination of the two approaches to improve forecasts of hydro-meteorological time series. WD allows decomposition and reconstruction of a time series by the wavelet transform, and hence separation of the noise from the original series. RSPA, a more reliable and efficient version of Set Pair Analysis, is integrated with WD to form the hybrid WD-RSPA approach. Two types of hydro-meteorological data sets with different characteristics and different levels of human influences at some representative stations are used to illustrate the WD-RSPA approach. The approach is also compared to three other generic methods: the conventional Auto Regressive Integrated Moving Average (ARIMA) method, Artificial Neural Networks (ANNs) (BP-error Back Propagation, MLP-Multilayer Perceptron and RBF-Radial Basis Function), and RSPA alone. Nine error metrics are used to evaluate the model performance. Compared to three other generic methods, the results generated by WD-REPA model presented invariably smaller error measures which means the forecasting capability of the WD-REPA model is better than other models. The results show that WD-RSPA is accurate, feasible, and effective. In particular, WD-RSPA is found to be the best among the various generic methods compared in this paper, even when the extreme events are included within a time series. Copyright © 2017 Elsevier Inc. All rights reserved.
Directory of Open Access Journals (Sweden)
Subanar Subanar
2006-01-01
Full Text Available Recently, one of the central topics for the neural networks (NN community is the issue of data preprocessing on the use of NN. In this paper, we will investigate this topic particularly on the effect of Decomposition method as data processing and the use of NN for modeling effectively time series with both trend and seasonal patterns. Limited empirical studies on seasonal time series forecasting with neural networks show that some find neural networks are able to model seasonality directly and prior deseasonalization is not necessary, and others conclude just the opposite. In this research, we study particularly on the effectiveness of data preprocessing, including detrending and deseasonalization by applying Decomposition method on NN modeling and forecasting performance. We use two kinds of data, simulation and real data. Simulation data are examined on multiplicative of trend and seasonality patterns. The results are compared to those obtained from the classical time series model. Our result shows that a combination of detrending and deseasonalization by applying Decomposition method is the effective data preprocessing on the use of NN for forecasting trend and seasonal time series.
Li, Shuying; Zhuang, Jun; Shen, Shifei
2017-07-01
In recent years, various types of terrorist attacks occurred, causing worldwide catastrophes. According to the Global Terrorism Database (GTD), among all attack tactics, bombing attacks happened most frequently, followed by armed assaults. In this article, a model for analyzing and forecasting the conditional probability of bombing attacks (CPBAs) based on time-series methods is developed. In addition, intervention analysis is used to analyze the sudden increase in the time-series process. The results show that the CPBA increased dramatically at the end of 2011. During that time, the CPBA increased by 16.0% in a two-month period to reach the peak value, but still stays 9.0% greater than the predicted level after the temporary effect gradually decays. By contrast, no significant fluctuation can be found in the conditional probability process of armed assault. It can be inferred that some social unrest, such as America's troop withdrawal from Afghanistan and Iraq, could have led to the increase of the CPBA in Afghanistan, Iraq, and Pakistan. The integrated time-series and intervention model is used to forecast the monthly CPBA in 2014 and through 2064. The average relative error compared with the real data in 2014 is 3.5%. The model is also applied to the total number of attacks recorded by the GTD between 2004 and 2014. © 2016 Society for Risk Analysis.
Applying Markov Chains for NDVI Time Series Forecasting of Latvian Regions
Directory of Open Access Journals (Sweden)
Stepchenko Arthur
2015-12-01
Full Text Available Time series of earth observation based estimates of vegetation inform about variations in vegetation at the scale of Latvia. A vegetation index is an indicator that describes the amount of chlorophyll (the green mass and shows the relative density and health of vegetation. NDVI index is an important variable for vegetation forecasting and management of various problems, such as climate change monitoring, energy usage monitoring, managing the consumption of natural resources, agricultural productivity monitoring, drought monitoring and forest fire detection. In this paper, we make a one-step-ahead prediction of 7-daily time series of NDVI index using Markov chains. The choice of a Markov chain is due to the fact that a Markov chain is a sequence of random variables where each variable is located in some state. And a Markov chain contains probabilities of moving from one state to other.
Big Data Mining of Energy Time Series for Behavioral Analytics and Energy Consumption Forecasting
Directory of Open Access Journals (Sweden)
Shailendra Singh
2018-02-01
Full Text Available Responsible, efficient and environmentally aware energy consumption behavior is becoming a necessity for the reliable modern electricity grid. In this paper, we present an intelligent data mining model to analyze, forecast and visualize energy time series to uncover various temporal energy consumption patterns. These patterns define the appliance usage in terms of association with time such as hour of the day, period of the day, weekday, week, month and season of the year as well as appliance-appliance associations in a household, which are key factors to infer and analyze the impact of consumers’ energy consumption behavior and energy forecasting trend. This is challenging since it is not trivial to determine the multiple relationships among different appliances usage from concurrent streams of data. Also, it is difficult to derive accurate relationships between interval-based events where multiple appliance usages persist for some duration. To overcome these challenges, we propose unsupervised data clustering and frequent pattern mining analysis on energy time series, and Bayesian network prediction for energy usage forecasting. We perform extensive experiments using real-world context-rich smart meter datasets. The accuracy results of identifying appliance usage patterns using the proposed model outperformed Support Vector Machine (SVM and Multi-Layer Perceptron (MLP at each stage while attaining a combined accuracy of 81.82%, 85.90%, 89.58% for 25%, 50% and 75% of the training data size respectively. Moreover, we achieved energy consumption forecast accuracies of 81.89% for short-term (hourly and 75.88%, 79.23%, 74.74%, and 72.81% for the long-term; i.e., day, week, month, and season respectively.
Time series modelling to forecast prehospital EMS demand for diabetic emergencies.
Villani, Melanie; Earnest, Arul; Nanayakkara, Natalie; Smith, Karen; de Courten, Barbora; Zoungas, Sophia
2017-05-05
Acute diabetic emergencies are often managed by prehospital Emergency Medical Services (EMS). The projected growth in prevalence of diabetes is likely to result in rising demand for prehospital EMS that are already under pressure. The aims of this study were to model the temporal trends and provide forecasts of prehospital attendances for diabetic emergencies. A time series analysis on monthly cases of hypoglycemia and hyperglycemia was conducted using data from the Ambulance Victoria (AV) electronic database between 2009 and 2015. Using the seasonal autoregressive integrated moving average (SARIMA) modelling process, different models were evaluated. The most parsimonious model with the highest accuracy was selected. Forty-one thousand four hundred fifty-four prehospital diabetic emergencies were attended over a seven-year period with an increase in the annual median monthly caseload between 2009 (484.5) and 2015 (549.5). Hypoglycemia (70%) and people with type 1 diabetes (48%) accounted for most attendances. The SARIMA (0,1,0,12) model provided the best fit, with a MAPE of 4.2% and predicts a monthly caseload of approximately 740 by the end of 2017. Prehospital EMS demand for diabetic emergencies is increasing. SARIMA time series models are a valuable tool to allow forecasting of future caseload with high accuracy and predict increasing cases of prehospital diabetic emergencies into the future. The model generated by this study may be used by service providers to allow appropriate planning and resource allocation of EMS for diabetic emergencies.
Grigoryeva, Lyudmila; Henriques, Julie; Larger, Laurent; Ortega, Juan-Pablo
2014-07-01
Reservoir computing is a recently introduced machine learning paradigm that has already shown excellent performances in the processing of empirical data. We study a particular kind of reservoir computers called time-delay reservoirs that are constructed out of the sampling of the solution of a time-delay differential equation and show their good performance in the forecasting of the conditional covariances associated to multivariate discrete-time nonlinear stochastic processes of VEC-GARCH type as well as in the prediction of factual daily market realized volatilities computed with intraday quotes, using as training input daily log-return series of moderate size. We tackle some problems associated to the lack of task-universality for individually operating reservoirs and propose a solution based on the use of parallel arrays of time-delay reservoirs. Copyright © 2014 Elsevier Ltd. All rights reserved.
A Time Series Model for Assessing the Trend and Forecasting the Road Traffic Accident Mortality.
Yousefzadeh-Chabok, Shahrokh; Ranjbar-Taklimie, Fatemeh; Malekpouri, Reza; Razzaghi, Alireza
2016-09-01
Road traffic accident (RTA) is one of the main causes of trauma and known as a growing public health concern worldwide, especially in developing countries. Assessing the trend of fatalities in the past years and forecasting it enables us to make the appropriate planning for prevention and control. This study aimed to assess the trend of RTAs and forecast it in the next years by using time series modeling. In this historical analytical study, the RTA mortalities in Zanjan Province, Iran, were evaluated during 2007 - 2013. The time series analyses including Box-Jenkins models were used to assess the trend of accident fatalities in previous years and forecast it for the next 4 years. The mean age of the victims was 37.22 years (SD = 20.01). From a total of 2571 deaths, 77.5% (n = 1992) were males and 22.5% (n = 579) were females. The study models showed a descending trend of fatalities in the study years. The SARIMA (1, 1, 3) (0, 1, 0) 12 model was recognized as a best fit model in forecasting the trend of fatalities. Forecasting model also showed a descending trend of traffic accident mortalities in the next 4 years. There was a decreasing trend in the study and the future years. It seems that implementation of some interventions in the recent decade has had a positive effect on the decline of RTA fatalities. Nevertheless, there is still a need to pay more attention in order to prevent the occurrence and the mortalities related to traffic accidents.
FPGA-Based Stochastic Echo State Networks for Time-Series Forecasting.
Alomar, Miquel L; Canals, Vincent; Perez-Mora, Nicolas; Martínez-Moll, Víctor; Rosselló, Josep L
2016-01-01
Hardware implementation of artificial neural networks (ANNs) allows exploiting the inherent parallelism of these systems. Nevertheless, they require a large amount of resources in terms of area and power dissipation. Recently, Reservoir Computing (RC) has arisen as a strategic technique to design recurrent neural networks (RNNs) with simple learning capabilities. In this work, we show a new approach to implement RC systems with digital gates. The proposed method is based on the use of probabilistic computing concepts to reduce the hardware required to implement different arithmetic operations. The result is the development of a highly functional system with low hardware resources. The presented methodology is applied to chaotic time-series forecasting.
Energy Technology Data Exchange (ETDEWEB)
Gallego, C. J.
2010-03-08
Abstract: This technical report is focused on the analysis of stochastic processes that switch between different dynamics (also called regimes or mechanisms) over time. The so-called Switching-regime models consider several underlying functions instead of one. In this case, a classification problem arises as the current regime has to be assessed at each time-step. The identification of the regimes allows the performance of regime-switching models for short-term forecasting purposes. Within this framework, identifying different regimes showed by time-series is the aim of this work. The proposed approach is based on a statistical tool called Gamma-test. One of the main advantages of this methodology is the absence of a mathematical definition for the different underlying functions. Applications with both simulated and real wind power data have been considered. Results on simulated time series show that regimes can be successfully identified under certain hypothesis. Nevertheless, this work highlights that further research has to be done when considering real wind power time-series, which usually show different behaviours (e.g. fluctuations or ramps, followed by low variance periods). A better understanding of these events eventually will improve wind power forecasting. (Author) 15 refs.
International Nuclear Information System (INIS)
Guo, Zhenhai; Chi, Dezhong; Wu, Jie; Zhang, Wenyu
2014-01-01
Highlights: • Impact of meteorological factors on wind speed forecasting is taken into account. • Forecasted wind speed results are corrected by the associated rules. • Forecasting accuracy is improved by the new wind speed forecasting strategy. • Robust of the proposed model is validated by data sampled from different sites. - Abstract: Wind energy has been the fastest growing renewable energy resource in recent years. Because of the intermittent nature of wind, wind power is a fluctuating source of electrical energy. Therefore, to minimize the impact of wind power on the electrical grid, accurate and reliable wind power forecasting is mandatory. In this paper, a new wind speed forecasting approach based on based on the chaotic time series modelling technique and the Apriori algorithm has been developed. The new approach consists of four procedures: (I) Clustering by using the k-means clustering approach; (II) Employing the Apriori algorithm to discover the association rules; (III) Forecasting the wind speed according to the chaotic time series forecasting model; and (IV) Correcting the forecasted wind speed data using the associated rules discovered previously. This procedure has been verified by 31-day-ahead daily average wind speed forecasting case studies, which employed the wind speed and other meteorological data collected from four meteorological stations located in the Hexi Corridor area of China. The results of these case studies reveal that the chaotic forecasting model can efficiently improve the accuracy of the wind speed forecasting, and the Apriori algorithm can effectively discover the association rules between the wind speed and other meteorological factors. In addition, the correction results demonstrate that the association rules discovered by the Apriori algorithm have powerful capacities in handling the forecasted wind speed values correction when the forecasted values do not match the classification discovered by the association rules
Financial Time Series Forecasting Using Directed-Weighted Chunking SVMs
Directory of Open Access Journals (Sweden)
Yongming Cai
2014-01-01
Full Text Available Support vector machines (SVMs are a promising alternative to traditional regression estimation approaches. But, when dealing with massive-scale data set, there exist many problems, such as the long training time and excessive demand of memory space. So, the SVMs algorithm is not suitable to deal with financial time series data. In order to solve these problems, directed-weighted chunking SVMs algorithm is proposed. In this algorithm, the whole training data set is split into several chunks, and then the support vectors are obtained on each subset. Furthermore, the weighted support vector regressions are calculated to obtain the forecast model on the new working data set. Our directed-weighted chunking algorithm provides a new method of support vectors decomposing and combining according to the importance of chunks, which can improve the operation speed without reducing prediction accuracy. Finally, IBM stock daily close prices data are used to verify the validity of the proposed algorithm.
A Time Series Model for Assessing the Trend and Forecasting the Road Traffic Accident Mortality
Yousefzadeh-Chabok, Shahrokh; Ranjbar-Taklimie, Fatemeh; Malekpouri, Reza; Razzaghi, Alireza
2016-01-01
Background Road traffic accident (RTA) is one of the main causes of trauma and known as a growing public health concern worldwide, especially in developing countries. Assessing the trend of fatalities in the past years and forecasting it enables us to make the appropriate planning for prevention and control. Objectives This study aimed to assess the trend of RTAs and forecast it in the next years by using time series modeling. Materials and Methods In this historical analytical study, the RTA mortalities in Zanjan Province, Iran, were evaluated during 2007 - 2013. The time series analyses including Box-Jenkins models were used to assess the trend of accident fatalities in previous years and forecast it for the next 4 years. Results The mean age of the victims was 37.22 years (SD = 20.01). From a total of 2571 deaths, 77.5% (n = 1992) were males and 22.5% (n = 579) were females. The study models showed a descending trend of fatalities in the study years. The SARIMA (1, 1, 3) (0, 1, 0) 12 model was recognized as a best fit model in forecasting the trend of fatalities. Forecasting model also showed a descending trend of traffic accident mortalities in the next 4 years. Conclusions There was a decreasing trend in the study and the future years. It seems that implementation of some interventions in the recent decade has had a positive effect on the decline of RTA fatalities. Nevertheless, there is still a need to pay more attention in order to prevent the occurrence and the mortalities related to traffic accidents. PMID:27800467
Jaber, Abobaker M; Ismail, Mohd Tahir; Altaher, Alsaidi M
2014-01-01
This paper mainly forecasts the daily closing price of stock markets. We propose a two-stage technique that combines the empirical mode decomposition (EMD) with nonparametric methods of local linear quantile (LLQ). We use the proposed technique, EMD-LLQ, to forecast two stock index time series. Detailed experiments are implemented for the proposed method, in which EMD-LPQ, EMD, and Holt-Winter methods are compared. The proposed EMD-LPQ model is determined to be superior to the EMD and Holt-Winter methods in predicting the stock closing prices.
FPGA-Based Stochastic Echo State Networks for Time-Series Forecasting
Directory of Open Access Journals (Sweden)
Miquel L. Alomar
2016-01-01
Full Text Available Hardware implementation of artificial neural networks (ANNs allows exploiting the inherent parallelism of these systems. Nevertheless, they require a large amount of resources in terms of area and power dissipation. Recently, Reservoir Computing (RC has arisen as a strategic technique to design recurrent neural networks (RNNs with simple learning capabilities. In this work, we show a new approach to implement RC systems with digital gates. The proposed method is based on the use of probabilistic computing concepts to reduce the hardware required to implement different arithmetic operations. The result is the development of a highly functional system with low hardware resources. The presented methodology is applied to chaotic time-series forecasting.
Earthquake forecasting studies using radon time series data in Taiwan
Walia, Vivek; Kumar, Arvind; Fu, Ching-Chou; Lin, Shih-Jung; Chou, Kuang-Wu; Wen, Kuo-Liang; Chen, Cheng-Hong
2017-04-01
For few decades, growing number of studies have shown usefulness of data in the field of seismogeochemistry interpreted as geochemical precursory signals for impending earthquakes and radon is idendified to be as one of the most reliable geochemical precursor. Radon is recognized as short-term precursor and is being monitored in many countries. This study is aimed at developing an effective earthquake forecasting system by inspecting long term radon time series data. The data is obtained from a network of radon monitoring stations eastblished along different faults of Taiwan. The continuous time series radon data for earthquake studies have been recorded and some significant variations associated with strong earthquakes have been observed. The data is also examined to evaluate earthquake precursory signals against environmental factors. An automated real-time database operating system has been developed recently to improve the data processing for earthquake precursory studies. In addition, the study is aimed at the appraisal and filtrations of these environmental parameters, in order to create a real-time database that helps our earthquake precursory study. In recent years, automatic operating real-time database has been developed using R, an open source programming language, to carry out statistical computation on the data. To integrate our data with our working procedure, we use the popular and famous open source web application solution, AMP (Apache, MySQL, and PHP), creating a website that could effectively show and help us manage the real-time database.
Time-Series Approaches for Forecasting the Number of Hospital Daily Discharged Inpatients.
Ting Zhu; Li Luo; Xinli Zhang; Yingkang Shi; Wenwu Shen
2017-03-01
For hospitals where decisions regarding acceptable rates of elective admissions are made in advance based on expected available bed capacity and emergency requests, accurate predictions of inpatient bed capacity are especially useful for capacity reservation purposes. As given, the remaining unoccupied beds at the end of each day, bed capacity of the next day can be obtained by examining the forecasts of the number of discharged patients during the next day. The features of fluctuations in daily discharges like trend, seasonal cycles, special-day effects, and autocorrelation complicate decision optimizing, while time-series models can capture these features well. This research compares three models: a model combining seasonal regression and ARIMA, a multiplicative seasonal ARIMA (MSARIMA) model, and a combinatorial model based on MSARIMA and weighted Markov Chain models in generating forecasts of daily discharges. The models are applied to three years of discharge data of an entire hospital. Several performance measures like the direction of the symmetry value, normalized mean squared error, and mean absolute percentage error are utilized to capture the under- and overprediction in model selection. The findings indicate that daily discharges can be forecast by using the proposed models. A number of important practical implications are discussed, such as the use of accurate forecasts in discharge planning, admission scheduling, and capacity reservation.
International Nuclear Information System (INIS)
Chai, Soo H.; Lim, Joon S.
2016-01-01
This study presents a forecasting model of cyclical fluctuations of the economy based on the time delay coordinate embedding method. The model uses a neuro-fuzzy network called neural network with weighted fuzzy membership functions (NEWFM). The preprocessed time series of the leading composite index using the time delay coordinate embedding method are used as input data to the NEWFM to forecast the business cycle. A comparative study is conducted using other methods based on wavelet transform and Principal Component Analysis for the performance comparison. The forecasting results are tested using a linear regression analysis to compare the approximation of the input data against the target class, gross domestic product (GDP). The chaos based model captures nonlinear dynamics and interactions within the system, which other two models ignore. The test results demonstrated that chaos based method significantly improved the prediction capability, thereby demonstrating superior performance to the other methods.
Directory of Open Access Journals (Sweden)
Juan Pardo
2015-04-01
Full Text Available Time series forecasting is an important predictive methodology which can be applied to a wide range of problems. Particularly, forecasting the indoor temperature permits an improved utilization of the HVAC (Heating, Ventilating and Air Conditioning systems in a home and thus a better energy efficiency. With such purpose the paper describes how to implement an Artificial Neural Network (ANN algorithm in a low cost system-on-chip to develop an autonomous intelligent wireless sensor network. The present paper uses a Wireless Sensor Networks (WSN to monitor and forecast the indoor temperature in a smart home, based on low resources and cost microcontroller technology as the 8051MCU. An on-line learning approach, based on Back-Propagation (BP algorithm for ANNs, has been developed for real-time time series learning. It performs the model training with every new data that arrive to the system, without saving enormous quantities of data to create a historical database as usual, i.e., without previous knowledge. Consequently to validate the approach a simulation study through a Bayesian baseline model have been tested in order to compare with a database of a real application aiming to see the performance and accuracy. The core of the paper is a new algorithm, based on the BP one, which has been described in detail, and the challenge was how to implement a computational demanding algorithm in a simple architecture with very few hardware resources.
Pardo, Juan; Zamora-Martínez, Francisco; Botella-Rocamora, Paloma
2015-01-01
Time series forecasting is an important predictive methodology which can be applied to a wide range of problems. Particularly, forecasting the indoor temperature permits an improved utilization of the HVAC (Heating, Ventilating and Air Conditioning) systems in a home and thus a better energy efficiency. With such purpose the paper describes how to implement an Artificial Neural Network (ANN) algorithm in a low cost system-on-chip to develop an autonomous intelligent wireless sensor network. The present paper uses a Wireless Sensor Networks (WSN) to monitor and forecast the indoor temperature in a smart home, based on low resources and cost microcontroller technology as the 8051MCU. An on-line learning approach, based on Back-Propagation (BP) algorithm for ANNs, has been developed for real-time time series learning. It performs the model training with every new data that arrive to the system, without saving enormous quantities of data to create a historical database as usual, i.e., without previous knowledge. Consequently to validate the approach a simulation study through a Bayesian baseline model have been tested in order to compare with a database of a real application aiming to see the performance and accuracy. The core of the paper is a new algorithm, based on the BP one, which has been described in detail, and the challenge was how to implement a computational demanding algorithm in a simple architecture with very few hardware resources. PMID:25905698
Pardo, Juan; Zamora-Martínez, Francisco; Botella-Rocamora, Paloma
2015-04-21
Time series forecasting is an important predictive methodology which can be applied to a wide range of problems. Particularly, forecasting the indoor temperature permits an improved utilization of the HVAC (Heating, Ventilating and Air Conditioning) systems in a home and thus a better energy efficiency. With such purpose the paper describes how to implement an Artificial Neural Network (ANN) algorithm in a low cost system-on-chip to develop an autonomous intelligent wireless sensor network. The present paper uses a Wireless Sensor Networks (WSN) to monitor and forecast the indoor temperature in a smart home, based on low resources and cost microcontroller technology as the 8051MCU. An on-line learning approach, based on Back-Propagation (BP) algorithm for ANNs, has been developed for real-time time series learning. It performs the model training with every new data that arrive to the system, without saving enormous quantities of data to create a historical database as usual, i.e., without previous knowledge. Consequently to validate the approach a simulation study through a Bayesian baseline model have been tested in order to compare with a database of a real application aiming to see the performance and accuracy. The core of the paper is a new algorithm, based on the BP one, which has been described in detail, and the challenge was how to implement a computational demanding algorithm in a simple architecture with very few hardware resources.
Time Series Analysis Forecasting and Control
Box, George E P; Reinsel, Gregory C
2011-01-01
A modernized new edition of one of the most trusted books on time series analysis. Since publication of the first edition in 1970, Time Series Analysis has served as one of the most influential and prominent works on the subject. This new edition maintains its balanced presentation of the tools for modeling and analyzing time series and also introduces the latest developments that have occurred n the field over the past decade through applications from areas such as business, finance, and engineering. The Fourth Edition provides a clearly written exploration of the key methods for building, cl
Forecasting long memory series subject to structural change: A two-stage approach
DEFF Research Database (Denmark)
Papailias, Fotis; Dias, Gustavo Fruet
2015-01-01
A two-stage forecasting approach for long memory time series is introduced. In the first step, we estimate the fractional exponent and, by applying the fractional differencing operator, obtain the underlying weakly dependent series. In the second step, we produce multi-step-ahead forecasts...... for the weakly dependent series and obtain their long memory counterparts by applying the fractional cumulation operator. The methodology applies to both stationary and nonstationary cases. Simulations and an application to seven time series provide evidence that the new methodology is more robust to structural...
Statistical models and time series forecasting of sulfur dioxide: a case study Tehran.
Hassanzadeh, S; Hosseinibalam, F; Alizadeh, R
2009-08-01
This study performed a time-series analysis, frequency distribution and prediction of SO(2) levels for five stations (Pardisan, Vila, Azadi, Gholhak and Bahman) in Tehran for the period of 2000-2005. Most sites show a quite similar characteristic with highest pollution in autumn-winter time and least pollution in spring-summer. The frequency distributions show higher peaks at two residential sites. The potential for SO(2) problems is high because of high emissions and the close geographical proximity of the major industrial and urban centers. The ACF and PACF are nonzero for several lags, indicating a mixed (ARMA) model, then at Bahman station an ARMA model was used for forecasting SO(2). The partial autocorrelations become close to 0 after about 5 lags while the autocorrelations remain strong through all the lags shown. The results proved that ARMA (2,2) model can provides reliable, satisfactory predictions for time series.
Time series modeling and forecasting using memetic algorithms for regime-switching models.
Bergmeir, Christoph; Triguero, Isaac; Molina, Daniel; Aznarte, José Luis; Benitez, José Manuel
2012-11-01
In this brief, we present a novel model fitting procedure for the neuro-coefficient smooth transition autoregressive model (NCSTAR), as presented by Medeiros and Veiga. The model is endowed with a statistically founded iterative building procedure and can be interpreted in terms of fuzzy rule-based systems. The interpretability of the generated models and a mathematically sound building procedure are two very important properties of forecasting models. The model fitting procedure employed by the original NCSTAR is a combination of initial parameter estimation by a grid search procedure with a traditional local search algorithm. We propose a different fitting procedure, using a memetic algorithm, in order to obtain more accurate models. An empirical evaluation of the method is performed, applying it to various real-world time series originating from three forecasting competitions. The results indicate that we can significantly enhance the accuracy of the models, making them competitive to models commonly used in the field.
High-Density Liquid-State Machine Circuitry for Time-Series Forecasting.
Rosselló, Josep L; Alomar, Miquel L; Morro, Antoni; Oliver, Antoni; Canals, Vincent
2016-08-01
Spiking neural networks (SNN) are the last neural network generation that try to mimic the real behavior of biological neurons. Although most research in this area is done through software applications, it is in hardware implementations in which the intrinsic parallelism of these computing systems are more efficiently exploited. Liquid state machines (LSM) have arisen as a strategic technique to implement recurrent designs of SNN with a simple learning methodology. In this work, we show a new low-cost methodology to implement high-density LSM by using Boolean gates. The proposed method is based on the use of probabilistic computing concepts to reduce hardware requirements, thus considerably increasing the neuron count per chip. The result is a highly functional system that is applied to high-speed time series forecasting.
Capan, Muge; Hoover, Stephen; Jackson, Eric V; Paul, David; Locke, Robert
2016-01-01
Accurate prediction of future patient census in hospital units is essential for patient safety, health outcomes, and resource planning. Forecasting census in the Neonatal Intensive Care Unit (NICU) is particularly challenging due to limited ability to control the census and clinical trajectories. The fixed average census approach, using average census from previous year, is a forecasting alternative used in clinical practice, but has limitations due to census variations. Our objectives are to: (i) analyze the daily NICU census at a single health care facility and develop census forecasting models, (ii) explore models with and without patient data characteristics obtained at the time of admission, and (iii) evaluate accuracy of the models compared with the fixed average census approach. We used five years of retrospective daily NICU census data for model development (January 2008 - December 2012, N=1827 observations) and one year of data for validation (January - December 2013, N=365 observations). Best-fitting models of ARIMA and linear regression were applied to various 7-day prediction periods and compared using error statistics. The census showed a slightly increasing linear trend. Best fitting models included a non-seasonal model, ARIMA(1,0,0), seasonal ARIMA models, ARIMA(1,0,0)x(1,1,2)7 and ARIMA(2,1,4)x(1,1,2)14, as well as a seasonal linear regression model. Proposed forecasting models resulted on average in 36.49% improvement in forecasting accuracy compared with the fixed average census approach. Time series models provide higher prediction accuracy under different census conditions compared with the fixed average census approach. Presented methodology is easily applicable in clinical practice, can be generalized to other care settings, support short- and long-term census forecasting, and inform staff resource planning.
Forecasting Hourly Water Demands With Seasonal Autoregressive Models for Real-Time Application
Chen, Jinduan; Boccelli, Dominic L.
2018-02-01
Consumer water demands are not typically measured at temporal or spatial scales adequate to support real-time decision making, and recent approaches for estimating unobserved demands using observed hydraulic measurements are generally not capable of forecasting demands and uncertainty information. While time series modeling has shown promise for representing total system demands, these models have generally not been evaluated at spatial scales appropriate for representative real-time modeling. This study investigates the use of a double-seasonal time series model to capture daily and weekly autocorrelations to both total system demands and regional aggregated demands at a scale that would capture demand variability across a distribution system. Emphasis was placed on the ability to forecast demands and quantify uncertainties with results compared to traditional time series pattern-based demand models as well as nonseasonal and single-seasonal time series models. Additional research included the implementation of an adaptive-parameter estimation scheme to update the time series model when unobserved changes occurred in the system. For two case studies, results showed that (1) for the smaller-scale aggregated water demands, the log-transformed time series model resulted in improved forecasts, (2) the double-seasonal model outperformed other models in terms of forecasting errors, and (3) the adaptive adjustment of parameters during forecasting improved the accuracy of the generated prediction intervals. These results illustrate the capabilities of time series modeling to forecast both water demands and uncertainty estimates at spatial scales commensurate for real-time modeling applications and provide a foundation for developing a real-time integrated demand-hydraulic model.
International Nuclear Information System (INIS)
Voyant, Cyril; Motte, Fabrice; Fouilloy, Alexis; Notton, Gilles; Paoli, Christophe; Nivet, Marie-Laure
2017-01-01
Integration of unpredictable renewable energy sources into electrical networks intensifies the complexity of the grid management due to their intermittent and unforeseeable nature. Because of the strong increase of solar power generation the prediction of solar yields becomes more and more important. Electrical operators need an estimation of the future production. For nowcasting and short term forecasting, the usual technics based on machine learning need large historical data sets of good quality during the training phase of predictors. However data are not always available and induce an advanced maintenance of meteorological stations, making the method inapplicable for poor instrumented or isolated sites. In this work, we propose intuitive methodologies based on the Kalman filter use (also known as linear quadratic estimation), able to predict a global radiation time series without the need of historical data. The accuracy of these methods is compared to other classical data driven methods, for different horizons of prediction and time steps. The proposed approach shows interesting capabilities allowing to improve quasi-systematically the prediction. For one to 10 h horizons Kalman model performances are competitive in comparison to more sophisticated models such as ANN which require both consistent historical data sets and computational resources. - Highlights: • Solar radiation forecasting with time series formalism. • Trainless approach compared to machine learning methods. • Very simple method dedicated to solar irradiation forecasting with high accuracy.
Simple nuclear norm based algorithms for imputing missing data and forecasting in time series
Butcher, Holly Louise; Gillard, Jonathan William
2017-01-01
There has been much recent progress on the use of the nuclear norm for the so-called matrix completion problem (the problem of imputing missing values of a matrix). In this paper we investigate the use of the nuclear norm for modelling time series, with particular attention to imputing missing data and forecasting. We introduce a simple alternating projections type algorithm based on the nuclear norm for these tasks, and consider a number of practical examples.
Ramli, Nazirah; Mutalib, Siti Musleha Ab; Mohamad, Daud
2017-08-01
Fuzzy time series forecasting model has been proposed since 1993 to cater for data in linguistic values. Many improvement and modification have been made to the model such as enhancement on the length of interval and types of fuzzy logical relation. However, most of the improvement models represent the linguistic term in the form of discrete fuzzy sets. In this paper, fuzzy time series model with data in the form of trapezoidal fuzzy numbers and natural partitioning length approach is introduced for predicting the unemployment rate. Two types of fuzzy relations are used in this study which are first order and second order fuzzy relation. This proposed model can produce the forecasted values under different degree of confidence.
International Nuclear Information System (INIS)
Voyant, Cyril; Notton, Gilles; Darras, Christophe; Fouilloy, Alexis; Motte, Fabrice
2017-01-01
As global solar radiation forecasting is a very important challenge, several methods are devoted to this goal with different levels of accuracy and confidence. In this study we propose to better understand how the uncertainty is propagated in the context of global radiation time series forecasting using machine learning. Indeed we propose to decompose the error considering four kinds of uncertainties: the error due to the measurement, the variability of time series, the machine learning uncertainty and the error related to the horizon. All these components of the error allow to determinate a global uncertainty generating prediction bands related to the prediction efficiency. We also have defined a reliability index which could be very interesting for the grid manager in order to estimate the validity of predictions. We have experimented this method on a multilayer perceptron which is a popular machine learning technique. We have shown that the global error and its components are essential to quantify in order to estimate the reliability of the model outputs. The described method has been successfully applied to four meteorological stations in Mediterranean area. - Highlights: • Solar irradiation predictions require confidence bands. • There are a lot of kinds of uncertainties to take into account in order to propose prediction bands. • the ranking of different kinds of uncertainties is essential to propose an operational tool for the grid managers.
Elements of nonlinear time series analysis and forecasting
De Gooijer, Jan G
2017-01-01
This book provides an overview of the current state-of-the-art of nonlinear time series analysis, richly illustrated with examples, pseudocode algorithms and real-world applications. Avoiding a “theorem-proof” format, it shows concrete applications on a variety of empirical time series. The book can be used in graduate courses in nonlinear time series and at the same time also includes interesting material for more advanced readers. Though it is largely self-contained, readers require an understanding of basic linear time series concepts, Markov chains and Monte Carlo simulation methods. The book covers time-domain and frequency-domain methods for the analysis of both univariate and multivariate (vector) time series. It makes a clear distinction between parametric models on the one hand, and semi- and nonparametric models/methods on the other. This offers the reader the option of concentrating exclusively on one of these nonlinear time series analysis methods. To make the book as user friendly as possible...
Foundations of Sequence-to-Sequence Modeling for Time Series
Kuznetsov, Vitaly; Mariet, Zelda
2018-01-01
The availability of large amounts of time series data, paired with the performance of deep-learning algorithms on a broad class of problems, has recently led to significant interest in the use of sequence-to-sequence models for time series forecasting. We provide the first theoretical analysis of this time series forecasting framework. We include a comparison of sequence-to-sequence modeling to classical time series models, and as such our theory can serve as a quantitative guide for practiti...
Hierarchical time series bottom-up approach for forecast the export value in Central Java
Mahkya, D. A.; Ulama, B. S.; Suhartono
2017-10-01
The purpose of this study is Getting the best modeling and predicting the export value of Central Java using a Hierarchical Time Series. The export value is one variable injection in the economy of a country, meaning that if the export value of the country increases, the country’s economy will increase even more. Therefore, it is necessary appropriate modeling to predict the export value especially in Central Java. Export Value in Central Java are grouped into 21 commodities with each commodity has a different pattern. One approach that can be used time series is a hierarchical approach. Hierarchical Time Series is used Buttom-up. To Forecast the individual series at all levels using Autoregressive Integrated Moving Average (ARIMA), Radial Basis Function Neural Network (RBFNN), and Hybrid ARIMA-RBFNN. For the selection of the best models used Symmetric Mean Absolute Percentage Error (sMAPE). Results of the analysis showed that for the Export Value of Central Java, Bottom-up approach with Hybrid ARIMA-RBFNN modeling can be used for long-term predictions. As for the short and medium-term predictions, it can be used a bottom-up approach RBFNN modeling. Overall bottom-up approach with RBFNN modeling give the best result.
Increasing the temporal resolution of direct normal solar irradiance forecasted series
Fernández-Peruchena, Carlos M.; Gastón, Martin; Schroedter-Homscheidt, Marion; Marco, Isabel Martínez; Casado-Rubio, José L.; García-Moya, José Antonio
2017-06-01
A detailed knowledge of the solar resource is a critical point in the design and control of Concentrating Solar Power (CSP) plants. In particular, accurate forecasting of solar irradiance is essential for the efficient operation of solar thermal power plants, the management of energy markets, and the widespread implementation of this technology. Numerical weather prediction (NWP) models are commonly used for solar radiation forecasting. In the ECMWF deterministic forecasting system, all forecast parameters are commercially available worldwide at 3-hourly intervals. Unfortunately, as Direct Normal solar Irradiance (DNI) exhibits a great variability due to the dynamic effects of passing clouds, 3-h time resolution is insufficient for accurate simulations of CSP plants due to their nonlinear response to DNI, governed by various thermal inertias due to their complex response characteristics. DNI series of hourly or sub-hourly frequency resolution are normally used for an accurate modeling and analysis of transient processes in CSP technologies. In this context, the objective of this study is to propose a methodology for generating synthetic DNI time series at 1-h (or higher) temporal resolution from 3-h DNI series. The methodology is based upon patterns as being defined with help of the clear-sky envelope approach together with a forecast of maximum DNI value, and it has been validated with high quality measured DNI data.
Real-time emergency forecasting technique for situation management systems
Kopytov, V. V.; Kharechkin, P. V.; Naumenko, V. V.; Tretyak, R. S.; Tebueva, F. B.
2018-05-01
The article describes the real-time emergency forecasting technique that allows increasing accuracy and reliability of forecasting results of any emergency computational model applied for decision making in situation management systems. Computational models are improved by the Improved Brown’s method applying fractal dimension to forecast short time series data being received from sensors and control systems. Reliability of emergency forecasting results is ensured by the invalid sensed data filtering according to the methods of correlation analysis.
Plazas-Nossa, Leonardo; Torres, Andrés
2014-01-01
The objective of this work is to introduce a forecasting method for UV-Vis spectrometry time series that combines principal component analysis (PCA) and discrete Fourier transform (DFT), and to compare the results obtained with those obtained by using DFT. Three time series for three different study sites were used: (i) Salitre wastewater treatment plant (WWTP) in Bogotá; (ii) Gibraltar pumping station in Bogotá; and (iii) San Fernando WWTP in Itagüí (in the south part of Medellín). Each of these time series had an equal number of samples (1051). In general terms, the results obtained are hardly generalizable, as they seem to be highly dependent on specific water system dynamics; however, some trends can be outlined: (i) for UV range, DFT and PCA/DFT forecasting accuracy were almost the same; (ii) for visible range, the PCA/DFT forecasting procedure proposed gives systematically lower forecasting errors and variability than those obtained with the DFT procedure; and (iii) for short forecasting times the PCA/DFT procedure proposed is more suitable than the DFT procedure, according to processing times obtained.
The Prediction of Teacher Turnover Employing Time Series Analysis.
Costa, Crist H.
The purpose of this study was to combine knowledge of teacher demographic data with time-series forecasting methods to predict teacher turnover. Moving averages and exponential smoothing were used to forecast discrete time series. The study used data collected from the 22 largest school districts in Iowa, designated as FACT schools. Predictions…
A new accuracy measure based on bounded relative error for time series forecasting.
Chen, Chao; Twycross, Jamie; Garibaldi, Jonathan M
2017-01-01
Many accuracy measures have been proposed in the past for time series forecasting comparisons. However, many of these measures suffer from one or more issues such as poor resistance to outliers and scale dependence. In this paper, while summarising commonly used accuracy measures, a special review is made on the symmetric mean absolute percentage error. Moreover, a new accuracy measure called the Unscaled Mean Bounded Relative Absolute Error (UMBRAE), which combines the best features of various alternative measures, is proposed to address the common issues of existing measures. A comparative evaluation on the proposed and related measures has been made with both synthetic and real-world data. The results indicate that the proposed measure, with user selectable benchmark, performs as well as or better than other measures on selected criteria. Though it has been commonly accepted that there is no single best accuracy measure, we suggest that UMBRAE could be a good choice to evaluate forecasting methods, especially for cases where measures based on geometric mean of relative errors, such as the geometric mean relative absolute error, are preferred.
Wang, Wen-Chuan; Chau, Kwok-Wing; Cheng, Chun-Tian; Qiu, Lin
2009-08-01
SummaryDeveloping a hydrological forecasting model based on past records is crucial to effective hydropower reservoir management and scheduling. Traditionally, time series analysis and modeling is used for building mathematical models to generate hydrologic records in hydrology and water resources. Artificial intelligence (AI), as a branch of computer science, is capable of analyzing long-series and large-scale hydrological data. In recent years, it is one of front issues to apply AI technology to the hydrological forecasting modeling. In this paper, autoregressive moving-average (ARMA) models, artificial neural networks (ANNs) approaches, adaptive neural-based fuzzy inference system (ANFIS) techniques, genetic programming (GP) models and support vector machine (SVM) method are examined using the long-term observations of monthly river flow discharges. The four quantitative standard statistical performance evaluation measures, the coefficient of correlation ( R), Nash-Sutcliffe efficiency coefficient ( E), root mean squared error (RMSE), mean absolute percentage error (MAPE), are employed to evaluate the performances of various models developed. Two case study river sites are also provided to illustrate their respective performances. The results indicate that the best performance can be obtained by ANFIS, GP and SVM, in terms of different evaluation criteria during the training and validation phases.
Forecast of electric power market to short-term: a time series approcah
International Nuclear Information System (INIS)
Costa, Roberio Neves Pelinca da.
1994-01-01
Three different time series approaches are analysed by this dissertation in the Brazilian electricity markert context. The aim is to compare the predictive performance of these approaches from a simulated exercise using the main series of the Brazilian consumption of electricity: Total Consumption, Industrial Consumption, Residencial Consumption and Commercial Consumption. One concludes that these appraches offer an enormous potentiality to the short-term planning system of the Electric Sector. Among the univariate models, the results for the analysed period point out that the forecast produced by Holt-Winter's models are more accurate than those produced by ARIMA and structural models. When explanatory variables are introduced in the last models, one can notice, in general, an improvement in the predictive performance of the models, although there is no sufficient evidence to consider that they are superior to Holt-Winter's models. The models with explanatory variables can be particularly useful, however, when one intends either to build scenarios or to study the effects of some variables on the consumption of electricity. (author). 73 refs., 19 figs., 13 tabs
The analysis of time series: an introduction
National Research Council Canada - National Science Library
Chatfield, Christopher
1989-01-01
.... A variety of practical examples are given to support the theory. The book covers a wide range of time-series topics, including probability models for time series, Box-Jenkins forecasting, spectral analysis, linear systems and system identification...
Gershon, Andrea; Thiruchelvam, Deva; Moineddin, Rahim; Zhao, Xiu Yan; Hwee, Jeremiah; To, Teresa
2017-06-01
Knowing trends in and forecasting hospitalization and emergency department visit rates for chronic obstructive pulmonary disease (COPD) can enable health care providers, hospitals, and health care decision makers to plan for the future. We conducted a time-series analysis using health care administrative data from the Province of Ontario, Canada, to determine previous trends in acute care hospitalization and emergency department visit rates for COPD and then to forecast future rates. Individuals aged 35 years and older with physician-diagnosed COPD were identified using four universal government health administrative databases and a validated case definition. Monthly COPD hospitalization and emergency department visit rates per 1,000 people with COPD were determined from 2003 to 2014 and then forecasted to 2024 using autoregressive integrated moving average models. Between 2003 and 2014, COPD prevalence increased from 8.9 to 11.1%. During that time, there were 274,951 hospitalizations and 290,482 emergency department visits for COPD. After accounting for seasonality, we found that monthly COPD hospitalization and emergency department visit rates per 1,000 individuals with COPD remained stable. COPD prevalence was forecasted to increase to 12.7% (95% confidence interval [CI], 11.4-14.1) by 2024, whereas monthly COPD hospitalization and emergency department visit rates per 1,000 people with COPD were forecasted to remain stable at 2.7 (95% CI, 1.6-4.4) and 3.7 (95% CI, 2.3-5.6), respectively. Forecasted age- and sex-stratified rates were also stable. COPD hospital and emergency department visit rates per 1,000 people with COPD have been stable for more than a decade and are projected to remain stable in the near future. Given increasing COPD prevalence, this means notably more COPD health service use in the future.
Empirical forecast of quiet time ionospheric Total Electron Content maps over Europe
Badeke, Ronny; Borries, Claudia; Hoque, Mainul M.; Minkwitz, David
2018-06-01
An accurate forecast of the atmospheric Total Electron Content (TEC) is helpful to investigate space weather influences on the ionosphere and technical applications like satellite-receiver radio links. The purpose of this work is to compare four empirical methods for a 24-h forecast of vertical TEC maps over Europe under geomagnetically quiet conditions. TEC map data are obtained from the Space Weather Application Center Ionosphere (SWACI) and the Universitat Politècnica de Catalunya (UPC). The time-series methods Standard Persistence Model (SPM), a 27 day median model (MediMod) and a Fourier Series Expansion are compared to maps for the entire year of 2015. As a representative of the climatological coefficient models the forecast performance of the Global Neustrelitz TEC model (NTCM-GL) is also investigated. Time periods of magnetic storms, which are identified with the Dst index, are excluded from the validation. By calculating the TEC values with the most recent maps, the time-series methods perform slightly better than the coefficient model NTCM-GL. The benefit of NTCM-GL is its independence on observational TEC data. Amongst the time-series methods mentioned, MediMod delivers the best overall performance regarding accuracy and data gap handling. Quiet-time SWACI maps can be forecasted accurately and in real-time by the MediMod time-series approach.
Predicting chaotic time series
International Nuclear Information System (INIS)
Farmer, J.D.; Sidorowich, J.J.
1987-01-01
We present a forecasting technique for chaotic data. After embedding a time series in a state space using delay coordinates, we ''learn'' the induced nonlinear mapping using local approximation. This allows us to make short-term predictions of the future behavior of a time series, using information based only on past values. We present an error estimate for this technique, and demonstrate its effectiveness by applying it to several examples, including data from the Mackey-Glass delay differential equation, Rayleigh-Benard convection, and Taylor-Couette flow
Energy Technology Data Exchange (ETDEWEB)
Lefieux, V
2007-10-15
Reseau de Transport d'Electricite (RTE), in charge of operating the French electric transportation grid, needs an accurate forecast of the power consumption in order to operate it correctly. The forecasts used everyday result from a model combining a nonlinear parametric regression and a SARIMA model. In order to obtain an adaptive forecasting model, nonparametric forecasting methods have already been tested without real success. In particular, it is known that a nonparametric predictor behaves badly with a great number of explanatory variables, what is commonly called the curse of dimensionality. Recently, semi parametric methods which improve the pure nonparametric approach have been proposed to estimate a regression function. Based on the concept of 'dimension reduction', one those methods (called MAVE : Moving Average -conditional- Variance Estimate) can apply to time series. We study empirically its effectiveness to predict the future values of an autoregressive time series. We then adapt this method, from a practical point of view, to forecast power consumption. We propose a partially linear semi parametric model, based on the MAVE method, which allows to take into account simultaneously the autoregressive aspect of the problem and the exogenous variables. The proposed estimation procedure is practically efficient. (author)
Initial results with time series forecasting of TJ-II heliac waveforms
International Nuclear Information System (INIS)
Farias, G.; Dormido-Canto, S.; Vega, J.; Díaz, N.
2015-01-01
This article discusses about how to apply forecasting techniques to predict future samples of plasma signals during a discharge. One application of the forecasting could be to detect in real time anomalous behaviors in fusion waveforms. The work describes the implementation of three prediction techniques; two of them based on machine learning methods such as artificial neural networks and support vector machines for regression. The results have shown that depending on the temporal horizon, the predictions match the real samples in most cases with an error less than 5%, even more the forecasting of five samples ahead can reach accuracy over 90% in most signals analyzed.
Robust Control Charts for Time Series Data
Croux, C.; Gelper, S.; Mahieu, K.
2010-01-01
This article presents a control chart for time series data, based on the one-step- ahead forecast errors of the Holt-Winters forecasting method. We use robust techniques to prevent that outliers affect the estimation of the control limits of the chart. Moreover, robustness is important to maintain
Directory of Open Access Journals (Sweden)
J D Velásquez
2012-06-01
Full Text Available Many time series with trend and seasonal pattern are successfully modeled and forecasted by the airline model of Box and Jenkins; however, this model neglects the presence of nonlinearity on data. In this paper, we propose a new nonlinear version of the airline model; for this, we replace the moving average linear component by a multilayer perceptron neural network. The proposedmodel is used for forecasting two benchmark time series; we found that theproposed model is able to forecast the time series with more accuracy that other traditional approaches.Muchas series de tiempo con tendencia y ciclos estacionales son exitosamente modeladas y pronosticadas usando el modelo airline de Box y Jenkins; sin embargo, la presencia de no linealidades en los datos son despreciadas por este modelo. En este artículo, se propone una nueva versión no lineal del modelo airline; para esto, se reemplaza la componente lineal de promedios móviles por un perceptrón multicapa. El modelo propuesto es usado para pronosticar dos series de tiempo benchmark; se encontró que el modelo propuesto es capaz de pronosticar las series de tiempo con mayor precisión que otras aproximaciones tradicionales.
Wang, K W; Deng, C; Li, J P; Zhang, Y Y; Li, X Y; Wu, M C
2017-04-01
Tuberculosis (TB) affects people globally and is being reconsidered as a serious public health problem in China. Reliable forecasting is useful for the prevention and control of TB. This study proposes a hybrid model combining autoregressive integrated moving average (ARIMA) with a nonlinear autoregressive (NAR) neural network for forecasting the incidence of TB from January 2007 to March 2016. Prediction performance was compared between the hybrid model and the ARIMA model. The best-fit hybrid model was combined with an ARIMA (3,1,0) × (0,1,1)12 and NAR neural network with four delays and 12 neurons in the hidden layer. The ARIMA-NAR hybrid model, which exhibited lower mean square error, mean absolute error, and mean absolute percentage error of 0·2209, 0·1373, and 0·0406, respectively, in the modelling performance, could produce more accurate forecasting of TB incidence compared to the ARIMA model. This study shows that developing and applying the ARIMA-NAR hybrid model is an effective method to fit the linear and nonlinear patterns of time-series data, and this model could be helpful in the prevention and control of TB.
International Nuclear Information System (INIS)
Zhong Jian; Dong Gang; Sun Yimei; Zhang Zhaoyang; Wu Yuqin
2016-01-01
The present work reports the development of nonlinear time series prediction method of genetic algorithm (GA) with singular spectrum analysis (SSA) for forecasting the surface wind of a point station in the South China Sea (SCS) with scatterometer observations. Before the nonlinear technique GA is used for forecasting the time series of surface wind, the SSA is applied to reduce the noise. The surface wind speed and surface wind components from scatterometer observations at three locations in the SCS have been used to develop and test the technique. The predictions have been compared with persistence forecasts in terms of root mean square error. The predicted surface wind with GA and SSA made up to four days (longer for some point station) in advance have been found to be significantly superior to those made by persistence model. This method can serve as a cost-effective alternate prediction technique for forecasting surface wind of a point station in the SCS basin. (paper)
Effectiveness of firefly algorithm based neural network in time series ...
African Journals Online (AJOL)
Effectiveness of firefly algorithm based neural network in time series forecasting. ... In the experiments, three well known time series were used to evaluate the performance. Results obtained were compared with ... Keywords: Time series, Artificial Neural Network, Firefly Algorithm, Particle Swarm Optimization, Overfitting ...
Track Irregularity Time Series Analysis and Trend Forecasting
Directory of Open Access Journals (Sweden)
Jia Chaolong
2012-01-01
Full Text Available The combination of linear and nonlinear methods is widely used in the prediction of time series data. This paper analyzes track irregularity time series data by using gray incidence degree models and methods of data transformation, trying to find the connotative relationship between the time series data. In this paper, GM (1,1 is based on first-order, single variable linear differential equations; after an adaptive improvement and error correction, it is used to predict the long-term changing trend of track irregularity at a fixed measuring point; the stochastic linear AR, Kalman filtering model, and artificial neural network model are applied to predict the short-term changing trend of track irregularity at unit section. Both long-term and short-term changes prove that the model is effective and can achieve the expected accuracy.
Directory of Open Access Journals (Sweden)
Yan Hong Chen
2016-01-01
Full Text Available This paper proposes a new electric load forecasting model by hybridizing the fuzzy time series (FTS and global harmony search algorithm (GHSA with least squares support vector machines (LSSVM, namely GHSA-FTS-LSSVM model. Firstly, the fuzzy c-means clustering (FCS algorithm is used to calculate the clustering center of each cluster. Secondly, the LSSVM is applied to model the resultant series, which is optimized by GHSA. Finally, a real-world example is adopted to test the performance of the proposed model. In this investigation, the proposed model is verified using experimental datasets from the Guangdong Province Industrial Development Database, and results are compared against autoregressive integrated moving average (ARIMA model and other algorithms hybridized with LSSVM including genetic algorithm (GA, particle swarm optimization (PSO, harmony search, and so on. The forecasting results indicate that the proposed GHSA-FTS-LSSVM model effectively generates more accurate predictive results.
A time series model: First-order integer-valued autoregressive (INAR(1))
Simarmata, D. M.; Novkaniza, F.; Widyaningsih, Y.
2017-07-01
Nonnegative integer-valued time series arises in many applications. A time series model: first-order Integer-valued AutoRegressive (INAR(1)) is constructed by binomial thinning operator to model nonnegative integer-valued time series. INAR (1) depends on one period from the process before. The parameter of the model can be estimated by Conditional Least Squares (CLS). Specification of INAR(1) is following the specification of (AR(1)). Forecasting in INAR(1) uses median or Bayesian forecasting methodology. Median forecasting methodology obtains integer s, which is cumulative density function (CDF) until s, is more than or equal to 0.5. Bayesian forecasting methodology forecasts h-step-ahead of generating the parameter of the model and parameter of innovation term using Adaptive Rejection Metropolis Sampling within Gibbs sampling (ARMS), then finding the least integer s, where CDF until s is more than or equal to u . u is a value taken from the Uniform(0,1) distribution. INAR(1) is applied on pneumonia case in Penjaringan, Jakarta Utara, January 2008 until April 2016 monthly.
Non-linear forecasting in high-frequency financial time series
Strozzi, F.; Zaldívar, J. M.
2005-08-01
A new methodology based on state space reconstruction techniques has been developed for trading in financial markets. The methodology has been tested using 18 high-frequency foreign exchange time series. The results are in apparent contradiction with the efficient market hypothesis which states that no profitable information about future movements can be obtained by studying the past prices series. In our (off-line) analysis positive gain may be obtained in all those series. The trading methodology is quite general and may be adapted to other financial time series. Finally, the steps for its on-line application are discussed.
Plazas-Nossa, Leonardo; Hofer, Thomas; Gruber, Günter; Torres, Andres
2017-02-01
This work proposes a methodology for the forecasting of online water quality data provided by UV-Vis spectrometry. Therefore, a combination of principal component analysis (PCA) to reduce the dimensionality of a data set and artificial neural networks (ANNs) for forecasting purposes was used. The results obtained were compared with those obtained by using discrete Fourier transform (DFT). The proposed methodology was applied to four absorbance time series data sets composed by a total number of 5705 UV-Vis spectra. Absolute percentage errors obtained by applying the proposed PCA/ANN methodology vary between 10% and 13% for all four study sites. In general terms, the results obtained were hardly generalizable, as they appeared to be highly dependent on specific dynamics of the water system; however, some trends can be outlined. PCA/ANN methodology gives better results than PCA/DFT forecasting procedure by using a specific spectra range for the following conditions: (i) for Salitre wastewater treatment plant (WWTP) (first hour) and Graz West R05 (first 18 min), from the last part of UV range to all visible range; (ii) for Gibraltar pumping station (first 6 min) for all UV-Vis absorbance spectra; and (iii) for San Fernando WWTP (first 24 min) for all of UV range to middle part of visible range.
Vlasenko, A. V.; Sizonenko, A. B.; Zhdanov, A. A.
2018-05-01
Discrete time series or mappings are proposed for describing the dynamics of a nonlinear system. The article considers the problems of forecasting the dynamics of the system from the time series generated by it. In particular, the commercial rate of drilling oil and gas wells can be considered as a series where each next value depends on the previous one. The main parameter here is the technical drilling speed. With the aim of eliminating the measurement error and presenting the commercial speed of the object to the current with a good accuracy, future or any of the elapsed time points, the use of the Kalman filter is suggested. For the transition from a deterministic model to a probabilistic one, the use of ensemble modeling is suggested. Ensemble systems can provide a wide range of visual output, which helps the user to evaluate the measure of confidence in the model. In particular, the availability of information on the estimated calendar duration of the construction of oil and gas wells will allow drilling companies to optimize production planning by rationalizing the approach to loading drilling rigs, which ultimately leads to maximization of profit and an increase of their competitiveness.
Cherednichenko, A. V.; Cherednichenko, A. V.; Cherednichenko, V. S.
2018-01-01
It is shown that a significant connection exists between the most important harmonics, extracted in the process of harmonic analysis of time series of precipitation in the catchment area of rivers and the amount of runoff. This allowed us to predict the size of the flow for a period of up to 20 years, assuming that the main parameters of the harmonics are preserved at the predicted time interval. The results of such a forecast for three river basins of Kazakhstan are presented.
Short time ahead wind power production forecast
International Nuclear Information System (INIS)
Sapronova, Alla; Meissner, Catherine; Mana, Matteo
2016-01-01
An accurate prediction of wind power output is crucial for efficient coordination of cooperative energy production from different sources. Long-time ahead prediction (from 6 to 24 hours) of wind power for onshore parks can be achieved by using a coupled model that would bridge the mesoscale weather prediction data and computational fluid dynamics. When a forecast for shorter time horizon (less than one hour ahead) is anticipated, an accuracy of a predictive model that utilizes hourly weather data is decreasing. That is because the higher frequency fluctuations of the wind speed are lost when data is averaged over an hour. Since the wind speed can vary up to 50% in magnitude over a period of 5 minutes, the higher frequency variations of wind speed and direction have to be taken into account for an accurate short-term ahead energy production forecast. In this work a new model for wind power production forecast 5- to 30-minutes ahead is presented. The model is based on machine learning techniques and categorization approach and using the historical park production time series and hourly numerical weather forecast. (paper)
Short time ahead wind power production forecast
Sapronova, Alla; Meissner, Catherine; Mana, Matteo
2016-09-01
An accurate prediction of wind power output is crucial for efficient coordination of cooperative energy production from different sources. Long-time ahead prediction (from 6 to 24 hours) of wind power for onshore parks can be achieved by using a coupled model that would bridge the mesoscale weather prediction data and computational fluid dynamics. When a forecast for shorter time horizon (less than one hour ahead) is anticipated, an accuracy of a predictive model that utilizes hourly weather data is decreasing. That is because the higher frequency fluctuations of the wind speed are lost when data is averaged over an hour. Since the wind speed can vary up to 50% in magnitude over a period of 5 minutes, the higher frequency variations of wind speed and direction have to be taken into account for an accurate short-term ahead energy production forecast. In this work a new model for wind power production forecast 5- to 30-minutes ahead is presented. The model is based on machine learning techniques and categorization approach and using the historical park production time series and hourly numerical weather forecast.
Time Series Forecasting with Missing Values
Shin-Fu Wu; Chia-Yung Chang; Shie-Jue Lee
2015-01-01
Time series prediction has become more popular in various kinds of applications such as weather prediction, control engineering, financial analysis, industrial monitoring, etc. To deal with real-world problems, we are often faced with missing values in the data due to sensor malfunctions or human errors. Traditionally, the missing values are simply omitted or replaced by means of imputation methods. However, omitting those missing values may cause temporal discontinuity. Imputation methods, o...
Abunama, Taher; Othman, Faridah
2017-06-01
Analysing the fluctuations of wastewater inflow rates in sewage treatment plants (STPs) is essential to guarantee a sufficient treatment of wastewater before discharging it to the environment. The main objectives of this study are to statistically analyze and forecast the wastewater inflow rates into the Bandar Tun Razak STP in Kuala Lumpur, Malaysia. A time series analysis of three years’ weekly influent data (156weeks) has been conducted using the Auto-Regressive Integrated Moving Average (ARIMA) model. Various combinations of ARIMA orders (p, d, q) have been tried to select the most fitted model, which was utilized to forecast the wastewater inflow rates. The linear regression analysis was applied to testify the correlation between the observed and predicted influents. ARIMA (3, 1, 3) model was selected with the highest significance R-square and lowest normalized Bayesian Information Criterion (BIC) value, and accordingly the wastewater inflow rates were forecasted to additional 52weeks. The linear regression analysis between the observed and predicted values of the wastewater inflow rates showed a positive linear correlation with a coefficient of 0.831.
Multiresolution analysis of Bursa Malaysia KLCI time series
Ismail, Mohd Tahir; Dghais, Amel Abdoullah Ahmed
2017-05-01
In general, a time series is simply a sequence of numbers collected at regular intervals over a period. Financial time series data processing is concerned with the theory and practice of processing asset price over time, such as currency, commodity data, and stock market data. The primary aim of this study is to understand the fundamental characteristics of selected financial time series by using the time as well as the frequency domain analysis. After that prediction can be executed for the desired system for in sample forecasting. In this study, multiresolution analysis which the assist of discrete wavelet transforms (DWT) and maximal overlap discrete wavelet transform (MODWT) will be used to pinpoint special characteristics of Bursa Malaysia KLCI (Kuala Lumpur Composite Index) daily closing prices and return values. In addition, further case study discussions include the modeling of Bursa Malaysia KLCI using linear ARIMA with wavelets to address how multiresolution approach improves fitting and forecasting results.
Essays on forecasting stationary and nonstationary economic time series
Bachmeier, Lance Joseph
This dissertation consists of three essays. Chapter II considers the question of whether M2 growth can be used to forecast inflation at horizons of up to ten years. A vector error correction (VEC) model serves as our benchmark model. We find that M2 growth does have marginal predictive content for inflation at horizons of more than two years, but only when allowing for cointegration and when the cointegrating rank and vector are specified a priori. When estimating the cointegration vector or failing to impose cointegration, there is no longer evidence of causality running from M2 growth to inflation at any forecast horizon. Finally, we present evidence that M2 needs to be redefined, as forecasts of the VEC model using data on M2 observed after 1993 are worse than the forecasts of an autoregressive model of inflation. Chapter III reconsiders the evidence for a "rockets and feathers" effect in gasoline markets. We estimate an error correction model of gasoline prices using daily data for the period 1985--1998 and fail to find any evidence of asymmetry. We show that previous work suffered from two problems. First, nonstationarity in some of the regressors was ignored, leading to invalid inference. Second, the weekly data used in previous work leads to a temporal aggregation problem, and thus biased estimates of impulse response functions. Chapter IV tests for a forecasting relationship between the volume of litigation and macroeconomic variables. We analyze annual data for the period 1960--2000 on the number of cases filed, real GDP, real consumption expenditures, inflation, unemployment, and interest rates. Bivariate Granger causality tests show that several of the macroeconomic variables can be used to forecast the volume of litigation, but show no evidence that the volume of litigation can be used to forecast any of the macroeconomic variables. The analysis is then extended to bivariate and multivariate regression models, and we find similar evidence to that of the
Koopman Operator Framework for Time Series Modeling and Analysis
Surana, Amit
2018-01-01
We propose an interdisciplinary framework for time series classification, forecasting, and anomaly detection by combining concepts from Koopman operator theory, machine learning, and linear systems and control theory. At the core of this framework is nonlinear dynamic generative modeling of time series using the Koopman operator which is an infinite-dimensional but linear operator. Rather than working with the underlying nonlinear model, we propose two simpler linear representations or model forms based on Koopman spectral properties. We show that these model forms are invariants of the generative model and can be readily identified directly from data using techniques for computing Koopman spectral properties without requiring the explicit knowledge of the generative model. We also introduce different notions of distances on the space of such model forms which is essential for model comparison/clustering. We employ the space of Koopman model forms equipped with distance in conjunction with classical machine learning techniques to develop a framework for automatic feature generation for time series classification. The forecasting/anomaly detection framework is based on using Koopman model forms along with classical linear systems and control approaches. We demonstrate the proposed framework for human activity classification, and for time series forecasting/anomaly detection in power grid application.
a novel two – factor high order fuzzy time series with applications to ...
African Journals Online (AJOL)
HOD
objectively with multiple – factor fuzzy time series, recurrent number of fuzzy relationships, and assigning weights to elements of fuzzy forecasting rules. In this paper, a novel two – factor high – order fuzzy time series forecasting method based on fuzzy C-means clustering and particle swarm optimization is proposed to ...
Using time series structural characteristics to analyze grain prices in food insecure countries
Davenport, Frank; Funk, Chris
2015-01-01
Two components of food security monitoring are accurate forecasts of local grain prices and the ability to identify unusual price behavior. We evaluated a method that can both facilitate forecasts of cross-country grain price data and identify dissimilarities in price behavior across multiple markets. This method, characteristic based clustering (CBC), identifies similarities in multiple time series based on structural characteristics in the data. Here, we conducted a simulation experiment to determine if CBC can be used to improve the accuracy of maize price forecasts. We then compared forecast accuracies among clustered and non-clustered price series over a rolling time horizon. We found that the accuracy of forecasts on clusters of time series were equal to or worse than forecasts based on individual time series. However, in the following experiment we found that CBC was still useful for price analysis. We used the clusters to explore the similarity of price behavior among Kenyan maize markets. We found that price behavior in the isolated markets of Mandera and Marsabit has become increasingly dissimilar from markets in other Kenyan cities, and that these dissimilarities could not be explained solely by geographic distance. The structural isolation of Mandera and Marsabit that we find in this paper is supported by field studies on food security and market integration in Kenya. Our results suggest that a market with a unique price series (as measured by structural characteristics that differ from neighboring markets) may lack market integration and food security.
Luo, Yi; Zhang, Tao; Li, Xiao-song
2016-05-01
To explore the application of fuzzy time series model based on fuzzy c-means clustering in forecasting monthly incidence of Hepatitis E in mainland China. Apredictive model (fuzzy time series method based on fuzzy c-means clustering) was developed using Hepatitis E incidence data in mainland China between January 2004 and July 2014. The incidence datafrom August 2014 to November 2014 were used to test the fitness of the predictive model. The forecasting results were compared with those resulted from traditional fuzzy time series models. The fuzzy time series model based on fuzzy c-means clustering had 0.001 1 mean squared error (MSE) of fitting and 6.977 5 x 10⁻⁴ MSE of forecasting, compared with 0.0017 and 0.0014 from the traditional forecasting model. The results indicate that the fuzzy time series model based on fuzzy c-means clustering has a better performance in forecasting incidence of Hepatitis E.
A travel time forecasting model based on change-point detection method
LI, Shupeng; GUANG, Xiaoping; QIAN, Yongsheng; ZENG, Junwei
2017-06-01
Travel time parameters obtained from road traffic sensors data play an important role in traffic management practice. A travel time forecasting model is proposed for urban road traffic sensors data based on the method of change-point detection in this paper. The first-order differential operation is used for preprocessing over the actual loop data; a change-point detection algorithm is designed to classify the sequence of large number of travel time data items into several patterns; then a travel time forecasting model is established based on autoregressive integrated moving average (ARIMA) model. By computer simulation, different control parameters are chosen for adaptive change point search for travel time series, which is divided into several sections of similar state.Then linear weight function is used to fit travel time sequence and to forecast travel time. The results show that the model has high accuracy in travel time forecasting.
Zhang, Yong; Zhong, Miner; Geng, Nana; Jiang, Yunjian
2017-01-01
The market demand for electric vehicles (EVs) has increased in recent years. Suitable models are necessary to understand and forecast EV sales. This study presents a singular spectrum analysis (SSA) as a univariate time-series model and vector autoregressive model (VAR) as a multivariate model. Empirical results suggest that SSA satisfactorily indicates the evolving trend and provides reasonable results. The VAR model, which comprised exogenous parameters related to the market on a monthly basis, can significantly improve the prediction accuracy. The EV sales in China, which are categorized into battery and plug-in EVs, are predicted in both short term (up to December 2017) and long term (up to 2020), as statistical proofs of the growth of the Chinese EV industry.
Juang, Wang-Chuan; Huang, Sin-Jhih; Huang, Fong-Dee; Cheng, Pei-Wen; Wann, Shue-Ren
2017-12-01
Emergency department (ED) overcrowding is acknowledged as an increasingly important issue worldwide. Hospital managers are increasingly paying attention to ED crowding in order to provide higher quality medical services to patients. One of the crucial elements for a good management strategy is demand forecasting. Our study sought to construct an adequate model and to forecast monthly ED visits. We retrospectively gathered monthly ED visits from January 2009 to December 2016 to carry out a time series autoregressive integrated moving average (ARIMA) analysis. Initial development of the model was based on past ED visits from 2009 to 2016. A best-fit model was further employed to forecast the monthly data of ED visits for the next year (2016). Finally, we evaluated the predicted accuracy of the identified model with the mean absolute percentage error (MAPE). The software packages SAS/ETS V.9.4 and Office Excel 2016 were used for all statistical analyses. A series of statistical tests showed that six models, including ARIMA (0, 0, 1), ARIMA (1, 0, 0), ARIMA (1, 0, 1), ARIMA (2, 0, 1), ARIMA (3, 0, 1) and ARIMA (5, 0, 1), were candidate models. The model that gave the minimum Akaike information criterion and Schwartz Bayesian criterion and followed the assumptions of residual independence was selected as the adequate model. Finally, a suitable ARIMA (0, 0, 1) structure, yielding a MAPE of 8.91%, was identified and obtained as Visit t =7111.161+(a t +0.37462 a t -1). The ARIMA (0, 0, 1) model can be considered adequate for predicting future ED visits, and its forecast results can be used to aid decision-making processes. © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2017. All rights reserved. No commercial use is permitted unless otherwise expressly granted.
Forecasting the Seasonal Timing of Maine's Lobster Fishery
Directory of Open Access Journals (Sweden)
Katherine E. Mills
2017-11-01
Full Text Available The fishery for American lobster is currently the highest-valued commercial fishery in the United States, worth over US$620 million in dockside value in 2015. During a marine heat wave in 2012, the fishery was disrupted by the early warming of spring ocean temperatures and subsequent influx of lobster landings. This situation resulted in a price collapse, as the supply chain was not prepared for the early and abundant landings of lobsters. Motivated by this series of events, we have developed a forecast of when the Maine (USA lobster fishery will shift into its high volume summer landings period. The forecast uses a regression approach to relate spring ocean temperatures derived from four NERACOOS buoys along the coast of Maine to the start day of the high landings period of the fishery. Tested against conditions in past years, the forecast is able to predict the start day to within 1 week of the actual start, and the forecast can be issued 3–4 months prior to the onset of the high-landings period, providing valuable lead-time for the fishery and its associated supply chain to prepare for the upcoming season. Forecast results are conveyed in a probabilistic manner and are updated weekly over a 6-week forecasting period so that users can assess the certainty and consistency of the forecast and factor the uncertainty into their use of the information in a given year. By focusing on the timing of events, this type of seasonal forecast provides climate-relevant information to users at time scales that are meaningful for operational decisions. As climate change alters seasonal phenology and reduces the reliability of past experience as a guide for future expectations, this type of forecast can enable fishing industry participants to better adjust to and prepare for operating in the context of climate change.
Time Series Outlier Detection Based on Sliding Window Prediction
Directory of Open Access Journals (Sweden)
Yufeng Yu
2014-01-01
Full Text Available In order to detect outliers in hydrological time series data for improving data quality and decision-making quality related to design, operation, and management of water resources, this research develops a time series outlier detection method for hydrologic data that can be used to identify data that deviate from historical patterns. The method first built a forecasting model on the history data and then used it to predict future values. Anomalies are assumed to take place if the observed values fall outside a given prediction confidence interval (PCI, which can be calculated by the predicted value and confidence coefficient. The use of PCI as threshold is mainly on the fact that it considers the uncertainty in the data series parameters in the forecasting model to address the suitable threshold selection problem. The method performs fast, incremental evaluation of data as it becomes available, scales to large quantities of data, and requires no preclassification of anomalies. Experiments with different hydrologic real-world time series showed that the proposed methods are fast and correctly identify abnormal data and can be used for hydrologic time series analysis.
Real-time forecasting of the April 11, 2012 Sumatra tsunami
Wang, Dailin; Becker, Nathan C.; Walsh, David; Fryer, Gerard J.; Weinstein, Stuart A.; McCreery, Charles S.; ,
2012-01-01
The April 11, 2012, magnitude 8.6 earthquake off the northern coast of Sumatra generated a tsunami that was recorded at sea-level stations as far as 4800 km from the epicenter and at four ocean bottom pressure sensors (DARTs) in the Indian Ocean. The governments of India, Indonesia, Sri Lanka, Thailand, and Maldives issued tsunami warnings for their coastlines. The United States' Pacific Tsunami Warning Center (PTWC) issued an Indian Ocean-wide Tsunami Watch Bulletin in its role as an Interim Service Provider for the region. Using an experimental real-time tsunami forecast model (RIFT), PTWC produced a series of tsunami forecasts during the event that were based on rapidly derived earthquake parameters, including initial location and Mwp magnitude estimates and the W-phase centroid moment tensor solutions (W-phase CMTs) obtained at PTWC and at the U. S. Geological Survey (USGS). We discuss the real-time forecast methodology and how successive, real-time tsunami forecasts using the latest W-phase CMT solutions improved the accuracy of the forecast.
Ng, Kar Yong; Awang, Norhashidah
2018-01-06
Frequent haze occurrences in Malaysia have made the management of PM 10 (particulate matter with aerodynamic less than 10 μm) pollution a critical task. This requires knowledge on factors associating with PM 10 variation and good forecast of PM 10 concentrations. Hence, this paper demonstrates the prediction of 1-day-ahead daily average PM 10 concentrations based on predictor variables including meteorological parameters and gaseous pollutants. Three different models were built. They were multiple linear regression (MLR) model with lagged predictor variables (MLR1), MLR model with lagged predictor variables and PM 10 concentrations (MLR2) and regression with time series error (RTSE) model. The findings revealed that humidity, temperature, wind speed, wind direction, carbon monoxide and ozone were the main factors explaining the PM 10 variation in Peninsular Malaysia. Comparison among the three models showed that MLR2 model was on a same level with RTSE model in terms of forecasting accuracy, while MLR1 model was the worst.
Fractal analysis and nonlinear forecasting of indoor 222Rn time series
International Nuclear Information System (INIS)
Pausch, G.; Bossew, P.; Hofmann, W.; Steger, F.
1998-01-01
Fractal analyses of indoor 222 Rn time series were performed using different chaos theory based measurements such as time delay method, Hurst's rescaled range analysis, capacity (fractal) dimension, and Lyapunov exponent. For all time series we calculated only positive Lyapunov exponents which is a hint to chaos, while the Hurst exponents were well below 0.5, indicating antipersistent behaviour (past trends tend to reverse in the future). These time series were also analyzed with a nonlinear prediction method which allowed an estimation of the embedding dimensions with some restrictions, limiting the prediction to about three relative time steps. (orig.)
Multi-granular trend detection for time-series analysis
van Goethem, A.I.; Staals, F.; Löffler, M.; Dykes, J.; Speckmann, B.
2017-01-01
Time series (such as stock prices) and ensembles (such as model runs for weather forecasts) are two important types of one-dimensional time-varying data. Such data is readily available in large quantities but visual analysis of the raw data quickly becomes infeasible, even for moderately sized data
Nonlinear Prediction Model for Hydrologic Time Series Based on Wavelet Decomposition
Kwon, H.; Khalil, A.; Brown, C.; Lall, U.; Ahn, H.; Moon, Y.
2005-12-01
Traditionally forecasting and characterizations of hydrologic systems is performed utilizing many techniques. Stochastic linear methods such as AR and ARIMA and nonlinear ones such as statistical learning theory based tools have been extensively used. The common difficulty to all methods is the determination of sufficient and necessary information and predictors for a successful prediction. Relationships between hydrologic variables are often highly nonlinear and interrelated across the temporal scale. A new hybrid approach is proposed for the simulation of hydrologic time series combining both the wavelet transform and the nonlinear model. The present model employs some merits of wavelet transform and nonlinear time series model. The Wavelet Transform is adopted to decompose a hydrologic nonlinear process into a set of mono-component signals, which are simulated by nonlinear model. The hybrid methodology is formulated in a manner to improve the accuracy of a long term forecasting. The proposed hybrid model yields much better results in terms of capturing and reproducing the time-frequency properties of the system at hand. Prediction results are promising when compared to traditional univariate time series models. An application of the plausibility of the proposed methodology is provided and the results conclude that wavelet based time series model can be utilized for simulating and forecasting of hydrologic variable reasonably well. This will ultimately serve the purpose of integrated water resources planning and management.
Multivariate Time Series Analysis for Optimum Production Forecast ...
African Journals Online (AJOL)
FIRST LADY
(Amstrong, 1994, Bates, 1969, Newbold and Granger, 1974 and Whinkler and Makridakis ... distinct, essential components of inventory management while the random production is first ... To achieve this goal, model parameters are estimated or ... important regression model in forecasting of that nature, hence this study.
Real Time Wave Forecasting Using Wind Time History and Genetic Programming
Directory of Open Access Journals (Sweden)
A.R. Kambekar
2014-12-01
Full Text Available The significant wave height and average wave period form an essential input for operational activities in ocean and coastal areas. Such information is important in issuing appropriate warnings to people planning any construction or instillation works in the oceanic environment. Many countries over the world routinely collect wave and wind data through a network of wave rider buoys. The data collecting agencies transmit the resulting information online to their registered users through an internet or a web-based system. Operational wave forecasts in addition to the measured data are also made and supplied online to the users. This paper discusses operational wave forecasting in real time mode at locations where wind rather than wave data are continuously recorded. It is based on the time series modeling and incorporates an artificial intelligence technique of genetic programming. The significant wave height and average wave period values are forecasted over a period of 96 hr in future from the observations of wind speed and directions extending to a similar time scale in the past. Wind measurements made by floating buoys at eight different locations around India over a period varying from 1.5 yr to 9.0 yr were considered. The platform of Matlab and C++ was used to develop a graphical user interface that will extend an internet based user-friendly access of the forecasts to any registered user of the data dissemination authority.
Forecasting of exported volume for brazilian fruits by time series analysis: an arima/garch approach
Directory of Open Access Journals (Sweden)
Abdinardo Moreira Barreto de Oliveira
2015-06-01
Full Text Available The aim of this paper was to offer econometric forecasting models to the Brazilian exported volume fruits, with a view to assisting the planning and production control, also motivated by the existence of a few published papers dealing with this issue. In this sense, it was used the ARIMA/GARCH models, considering, likewise, the occurrence of a multiplicative stochastic seasonality in these series. They were collected 300 observations of exported net weight (kg between Jan/1989 and Dec/2013 of the following fruits: pineapple, banana, orange, lemon, apple, papaya, mango, watermelon, melon and grape, which selection criteria was its importance in the exported basket fruit, because they represented 97% of total received dollars, and 99% of total volume sold in 2010, of a population about 28 kinds of exported fruits. The results showed that it was not only observed the existence of a 12 month multiplicative seasonality in banana and mango. On the other hand, they were identified two fruits groups: (1 those which are continuously exported, and (2 those which have export peaks. On the quality of the models, they were considered satisfactory for six of the ten fruits analyzed. On the volatility, it was seen a high persistence in banana and papaya series, pointing to the existence of a structural break in time series, which could be linked to the economic crises happened in the last 17 years.
Forecasting Rubber Production Using Intelligent Time Series Analysis to Support Decision Makers
Subsorn, Panida; Xiao, Jitian; Clayden, Judy
2010-01-01
This chapter has investigated the best-fitting forecasting model for national rubber production forecasting for 2007 and 2008. The methods used in this study were based on non-neural network training and neural network training techniques to compare with the actual rubber production data for the best-fitting forecasting model. Hence, neural network training was presented to obtain more accurate forecasts for 2007 and 2008. To our knowledge, this is the preliminary study that brings a new pers...
Directory of Open Access Journals (Sweden)
Rui Xue
2015-01-01
Full Text Available Although bus passenger demand prediction has attracted increased attention during recent years, limited research has been conducted in the context of short-term passenger demand forecasting. This paper proposes an interactive multiple model (IMM filter algorithm-based model to predict short-term passenger demand. After aggregated in 15 min interval, passenger demand data collected from a busy bus route over four months were used to generate time series. Considering that passenger demand exhibits various characteristics in different time scales, three time series were developed, named weekly, daily, and 15 min time series. After the correlation, periodicity, and stationarity analyses, time series models were constructed. Particularly, the heteroscedasticity of time series was explored to achieve better prediction performance. Finally, IMM filter algorithm was applied to combine individual forecasting models with dynamically predicted passenger demand for next interval. Different error indices were adopted for the analyses of individual and hybrid models. The performance comparison indicates that hybrid model forecasts are superior to individual ones in accuracy. Findings of this study are of theoretical and practical significance in bus scheduling.
Amigó, José M; Hirata, Yoshito; Aihara, Kazuyuki
2017-08-01
In a previous paper, the authors studied the limits of probabilistic prediction in nonlinear time series analysis in a perfect model scenario, i.e., in the ideal case that the uncertainty of an otherwise deterministic model is due to only the finite precision of the observations. The model consisted of the symbolic dynamics of a measure-preserving transformation with respect to a finite partition of the state space, and the quality of the predictions was measured by the so-called ignorance score, which is a conditional entropy. In practice, though, partitions are dispensed with by considering numerical and experimental data to be continuous, which prompts us to trade off in this paper the Shannon entropy for the differential entropy. Despite technical differences, we show that the core of the previous results also hold in this extended scenario for sufficiently high precision. The corresponding imperfect model scenario will be revisited too because it is relevant for the applications. The theoretical part and its application to probabilistic forecasting are illustrated with numerical simulations and a new prediction algorithm.
International Nuclear Information System (INIS)
Burr, T.L.
1994-04-01
This report is a primer on the analysis of both linear and nonlinear time series with applications in nuclear safeguards and nonproliferation. We analyze eight simulated and two real time series using both linear and nonlinear modeling techniques. The theoretical treatment is brief but references to pertinent theory are provided. Forecasting is our main goal. However, because our most common approach is to fit models to the data, we also emphasize checking model adequacy by analyzing forecast errors for serial correlation or nonconstant variance
Forecasting Natural Gas Prices Using Wavelets, Time Series, and Artificial Neural Networks.
Jin, Junghwan; Kim, Jinsoo
2015-01-01
Following the unconventional gas revolution, the forecasting of natural gas prices has become increasingly important because the association of these prices with those of crude oil has weakened. With this as motivation, we propose some modified hybrid models in which various combinations of the wavelet approximation, detail components, autoregressive integrated moving average, generalized autoregressive conditional heteroskedasticity, and artificial neural network models are employed to predict natural gas prices. We also emphasize the boundary problem in wavelet decomposition, and compare results that consider the boundary problem case with those that do not. The empirical results show that our suggested approach can handle the boundary problem, such that it facilitates the extraction of the appropriate forecasting results. The performance of the wavelet-hybrid approach was superior in all cases, whereas the application of detail components in the forecasting was only able to yield a small improvement in forecasting performance. Therefore, forecasting with only an approximation component would be acceptable, in consideration of forecasting efficiency.
Forecasting Natural Gas Prices Using Wavelets, Time Series, and Artificial Neural Networks.
Directory of Open Access Journals (Sweden)
Junghwan Jin
Full Text Available Following the unconventional gas revolution, the forecasting of natural gas prices has become increasingly important because the association of these prices with those of crude oil has weakened. With this as motivation, we propose some modified hybrid models in which various combinations of the wavelet approximation, detail components, autoregressive integrated moving average, generalized autoregressive conditional heteroskedasticity, and artificial neural network models are employed to predict natural gas prices. We also emphasize the boundary problem in wavelet decomposition, and compare results that consider the boundary problem case with those that do not. The empirical results show that our suggested approach can handle the boundary problem, such that it facilitates the extraction of the appropriate forecasting results. The performance of the wavelet-hybrid approach was superior in all cases, whereas the application of detail components in the forecasting was only able to yield a small improvement in forecasting performance. Therefore, forecasting with only an approximation component would be acceptable, in consideration of forecasting efficiency.
Nonlinear Time Series Prediction Using Chaotic Neural Networks
Li, Ke-Ping; Chen, Tian-Lun
2001-06-01
A nonlinear feedback term is introduced into the evaluation equation of weights of the backpropagation algorithm for neural network, the network becomes a chaotic one. For the purpose of that we can investigate how the different feedback terms affect the process of learning and forecasting, we use the model to forecast the nonlinear time series which is produced by Makey-Glass equation. By selecting the suitable feedback term, the system can escape from the local minima and converge to the global minimum or its approximate solutions, and the forecasting results are better than those of backpropagation algorithm. The project supported by National Basic Research Project "Nonlinear Science" and National Natural Science Foundation of China under Grant No. 60074020
A Comparison of Various Forecasting Methods for Autocorrelated Time Series
Directory of Open Access Journals (Sweden)
Karin Kandananond
2012-07-01
Full Text Available The accuracy of forecasts significantly affects the overall performance of a whole supply chain system. Sometimes, the nature of consumer products might cause difficulties in forecasting for the future demands because of its complicated structure. In this study, two machine learning methods, artificial neural network (ANN and support vector machine (SVM, and a traditional approach, the autoregressive integrated moving average (ARIMA model, were utilized to predict the demand for consumer products. The training data used were the actual demand of six different products from a consumer product company in Thailand. Initially, each set of data was analysed using Ljung‐Box‐Q statistics to test for autocorrelation. Afterwards, each method was applied to different sets of data. The results indicated that the SVM method had a better forecast quality (in terms of MAPE than ANN and ARIMA in every category of products.
Time-series prediction and applications a machine intelligence approach
Konar, Amit
2017-01-01
This book presents machine learning and type-2 fuzzy sets for the prediction of time-series with a particular focus on business forecasting applications. It also proposes new uncertainty management techniques in an economic time-series using type-2 fuzzy sets for prediction of the time-series at a given time point from its preceding value in fluctuating business environments. It employs machine learning to determine repetitively occurring similar structural patterns in the time-series and uses stochastic automaton to predict the most probabilistic structure at a given partition of the time-series. Such predictions help in determining probabilistic moves in a stock index time-series Primarily written for graduate students and researchers in computer science, the book is equally useful for researchers/professionals in business intelligence and stock index prediction. A background of undergraduate level mathematics is presumed, although not mandatory, for most of the sections. Exercises with tips are provided at...
Kusev, Petko; van Schaik, Paul; Tsaneva-Atanasova, Krasimira; Juliusson, Asgeir; Chater, Nick
2018-01-01
When attempting to predict future events, people commonly rely on historical data. One psychological characteristic of judgmental forecasting of time series, established by research, is that when people make forecasts from series, they tend to underestimate future values for upward trends and overestimate them for downward ones, so-called trend-damping (modeled by anchoring on, and insufficient adjustment from, the average of recent time series values). Events in a time series can be experienced sequentially (dynamic mode), or they can also be retrospectively viewed simultaneously (static mode), not experienced individually in real time. In one experiment, we studied the influence of presentation mode (dynamic and static) on two sorts of judgment: (a) predictions of the next event (forecast) and (b) estimation of the average value of all the events in the presented series (average estimation). Participants' responses in dynamic mode were anchored on more recent events than in static mode for all types of judgment but with different consequences; hence, dynamic presentation improved prediction accuracy, but not estimation. These results are not anticipated by existing theoretical accounts; we develop and present an agent-based model-the adaptive anchoring model (ADAM)-to account for the difference between processing sequences of dynamically and statically presented stimuli (visually presented data). ADAM captures how variation in presentation mode produces variation in responses (and the accuracy of these responses) in both forecasting and judgment tasks. ADAM's model predictions for the forecasting and judgment tasks fit better with the response data than a linear-regression time series model. Moreover, ADAM outperformed autoregressive-integrated-moving-average (ARIMA) and exponential-smoothing models, while neither of these models accounts for people's responses on the average estimation task. Copyright © 2017 The Authors. Cognitive Science published by Wiley
Wangdi, Kinley; Singhasivanon, Pratap; Silawan, Tassanee; Lawpoolsri, Saranath; White, Nicholas J; Kaewkungwal, Jaranit
2010-09-03
Malaria still remains a public health problem in some districts of Bhutan despite marked reduction of cases in last few years. To strengthen the country's prevention and control measures, this study was carried out to develop forecasting and prediction models of malaria incidence in the endemic districts of Bhutan using time series and ARIMAX. This study was carried out retrospectively using the monthly reported malaria cases from the health centres to Vector-borne Disease Control Programme (VDCP) and the meteorological data from Meteorological Unit, Department of Energy, Ministry of Economic Affairs. Time series analysis was performed on monthly malaria cases, from 1994 to 2008, in seven malaria endemic districts. The time series models derived from a multiplicative seasonal autoregressive integrated moving average (ARIMA) was deployed to identify the best model using data from 1994 to 2006. The best-fit model was selected for each individual district and for the overall endemic area was developed and the monthly cases from January to December 2009 and 2010 were forecasted. In developing the prediction model, the monthly reported malaria cases and the meteorological factors from 1996 to 2008 of the seven districts were analysed. The method of ARIMAX modelling was employed to determine predictors of malaria of the subsequent month. It was found that the ARIMA (p, d, q) (P, D, Q)s model (p and P representing the auto regressive and seasonal autoregressive; d and D representing the non-seasonal differences and seasonal differencing; and q and Q the moving average parameters and seasonal moving average parameters, respectively and s representing the length of the seasonal period) for the overall endemic districts was (2,1,1)(0,1,1)12; the modelling data from each district revealed two most common ARIMA models including (2,1,1)(0,1,1)12 and (1,1,1)(0,1,1)12. The forecasted monthly malaria cases from January to December 2009 and 2010 varied from 15 to 82 cases in 2009
Monthly electric energy demand forecasting with neural networks and Fourier series
International Nuclear Information System (INIS)
Gonzalez-Romera, E.; Jaramillo-Moran, M.A.; Carmona-Fernandez, D.
2008-01-01
Medium-term electric energy demand forecasting is a useful tool for grid maintenance planning and market research of electric energy companies. Several methods, such as ARIMA, regression or artificial intelligence, have been usually used to carry out those predictions. Some approaches include weather or economic variables, which strongly influence electric energy demand. Economic variables usually influence the general series trend, while weather provides a periodic behavior because of its seasonal nature. This work investigates the periodic behavior of the Spanish monthly electric demand series, obtained by rejecting the trend from the consumption series. A novel hybrid approach is proposed: the periodic behavior is forecasted with a Fourier series while the trend is predicted with a neural network. Satisfactory results have been obtained, with a lower than 2% MAPE, which improve those reached when only neural networks or ARIMA were used for the same purpose. (author)
Forecasting ocean wave energy: A Comparison of the ECMWF wave model with time series methods
DEFF Research Database (Denmark)
Reikard, Gordon; Pinson, Pierre; Bidlot, Jean
2011-01-01
Recently, the technology has been developed to make wave farms commercially viable. Since electricity is perishable, utilities will be interested in forecasting ocean wave energy. The horizons involved in short-term management of power grids range from as little as a few hours to as long as several...... days. In selecting a method, the forecaster has a choice between physics-based models and statistical techniques. A further idea is to combine both types of models. This paper analyzes the forecasting properties of a well-known physics-based model, the European Center for Medium-Range Weather Forecasts...... (ECMWF) Wave Model, and two statistical techniques, time-varying parameter regressions and neural networks. Thirteen data sets at locations in the Atlantic and Pacific Oceans and the Gulf of Mexico are tested. The quantities to be predicted are the significant wave height, the wave period, and the wave...
Time-series modeling: applications to long-term finfish monitoring data
International Nuclear Information System (INIS)
Bireley, L.E.
1985-01-01
The growing concern and awareness that developed during the 1970's over the effects that industry had on the environment caused the electric utility industry in particular to develop monitoring programs. These programs generate long-term series of data that are not very amenable to classical normal-theory statistical analysis. The monitoring data collected from three finfish programs (impingement, trawl and seine) at the Millstone Nuclear Power Station were typical of such series and thus were used to develop methodology that used the full extent of the information in the series. The basis of the methodology was classic Box-Jenkins time-series modeling; however, the models also included deterministic components that involved flow, season and time as predictor variables. Time entered into the models as harmonic regression terms. Of the 32 models fitted to finfish catch data, 19 were found to account for more than 70% of the historical variation. The models were than used to forecast finfish catches a year in advance and comparisons were made to actual data. Usually the confidence intervals associated with the forecasts encompassed most of the observed data. The technique can provide the basis for intervention analysis in future impact assessments
Data-driven strategies for robust forecast of continuous glucose monitoring time-series.
Fiorini, Samuele; Martini, Chiara; Malpassi, Davide; Cordera, Renzo; Maggi, Davide; Verri, Alessandro; Barla, Annalisa
2017-07-01
Over the past decade, continuous glucose monitoring (CGM) has proven to be a very resourceful tool for diabetes management. To date, CGM devices are employed for both retrospective and online applications. Their use allows to better describe the patients' pathology as well as to achieve a better control of patients' level of glycemia. The analysis of CGM sensor data makes possible to observe a wide range of metrics, such as the glycemic variability during the day or the amount of time spent below or above certain glycemic thresholds. However, due to the high variability of the glycemic signals among sensors and individuals, CGM data analysis is a non-trivial task. Standard signal filtering solutions fall short when an appropriate model personalization is not applied. State-of-the-art data-driven strategies for online CGM forecasting rely upon the use of recursive filters. Each time a new sample is collected, such models need to adjust their parameters in order to predict the next glycemic level. In this paper we aim at demonstrating that the problem of online CGM forecasting can be successfully tackled by personalized machine learning models, that do not need to recursively update their parameters.
Directory of Open Access Journals (Sweden)
Yuqi Dong
2016-12-01
Full Text Available Accurate short-term electrical load forecasting plays a pivotal role in the national economy and people’s livelihood through providing effective future plans and ensuring a reliable supply of sustainable electricity. Although considerable work has been done to select suitable models and optimize the model parameters to forecast the short-term electrical load, few models are built based on the characteristics of time series, which will have a great impact on the forecasting accuracy. For that reason, this paper proposes a hybrid model based on data decomposition considering periodicity, trend and randomness of the original electrical load time series data. Through preprocessing and analyzing the original time series, the generalized regression neural network optimized by genetic algorithm is used to forecast the short-term electrical load. The experimental results demonstrate that the proposed hybrid model can not only achieve a good fitting ability, but it can also approximate the actual values when dealing with non-linear time series data with periodicity, trend and randomness.
Mutual Information-Based Inputs Selection for Electric Load Time Series Forecasting
Directory of Open Access Journals (Sweden)
Nenad Floranović
2013-02-01
Full Text Available Providing accurate load forecast to electric utility corporations is essential in order to reduce their operational costs and increase profits. Hence, training set selection is an important preprocessing step which has to be considered in practice in order to increase the accuracy of load forecasts. The usage of mutual information (MI has been recently proposed in regression tasks, mostly for feature selection and for identifying the real instances from training sets that contains noise and outliers. This paper proposes a methodology for the training set selection in a least squares support vector machines (LS-SVMs load forecasting model. A new application of the concept of MI is presented for the selection of a training set based on MI computation between initial training set instances and testing set instances. Accordingly, several LS-SVMs models have been trained, based on the proposed methodology, for hourly prediction of electric load for one day ahead. The results obtained from a real-world data set indicate that the proposed method increases the accuracy of load forecasting as well as reduces the size of the initial training set needed for model training.
RADON CONCENTRATION TIME SERIES MODELING AND APPLICATION DISCUSSION.
Stránský, V; Thinová, L
2017-11-01
In the year 2010 a continual radon measurement was established at Mladeč Caves in the Czech Republic using a continual radon monitor RADIM3A. In order to model radon time series in the years 2010-15, the Box-Jenkins Methodology, often used in econometrics, was applied. Because of the behavior of radon concentrations (RCs), a seasonal integrated, autoregressive moving averages model with exogenous variables (SARIMAX) has been chosen to model the measured time series. This model uses the time series seasonality, previously acquired values and delayed atmospheric parameters, to forecast RC. The developed model for RC time series is called regARIMA(5,1,3). Model residuals could be retrospectively compared with seismic evidence of local or global earthquakes, which occurred during the RCs measurement. This technique enables us to asses if continuously measured RC could serve an earthquake precursor. © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com.
time-series analysis of nigeria rice supply and demand: error ...
African Journals Online (AJOL)
O. Ojogho Ph.D
the two series have been changing which may continue for longer time than foreseen {Figure (1c)}. Figure (c) shows a forecast of rice supply and demand for Nigeria. It shows that beyond 2010, rice supply will permanently lead rice demand. This indicates that they either have time-varying means, time-varying variances or ...
State-space prediction model for chaotic time series
Alparslan, A. K.; Sayar, M.; Atilgan, A. R.
1998-08-01
A simple method for predicting the continuation of scalar chaotic time series ahead in time is proposed. The false nearest neighbors technique in connection with the time-delayed embedding is employed so as to reconstruct the state space. A local forecasting model based upon the time evolution of the topological neighboring in the reconstructed phase space is suggested. A moving root-mean-square error is utilized in order to monitor the error along the prediction horizon. The model is tested for the convection amplitude of the Lorenz model. The results indicate that for approximately 100 cycles of the training data, the prediction follows the actual continuation very closely about six cycles. The proposed model, like other state-space forecasting models, captures the long-term behavior of the system due to the use of spatial neighbors in the state space.
Directory of Open Access Journals (Sweden)
Hossein Hosseinpour Niknam
2012-07-01
Full Text Available In this research in order to forecast drought for the next coming year in Zahedan, using previous Standardized Precipitation Index (SPI data and 19 other climate indices were used. For this purpose Adaptive Neuro-Fuzzy Inference System (ANFIS was applied to build the predicting model and SPI drought index for drought quantity. At first calculating correlation approach for analysis between droughts and climate indices was used and the most suitable indices were selected. In the next stage drought prediction for period of 12 months was done. Different combinations among input variables in ANFIS models were entered. SPI drought index was the output of the model. The results showed that just using time series like the previous year drought SPI index in forecasting the 12 month drought was effective. However among all climate indices that were used, Nino4 showed the most suitable results.
DEFF Research Database (Denmark)
Pinson, Pierre; Madsen, Henrik
2008-01-01
Better modelling and forecasting of very short-term power fluctuations at large offshore wind farms may significantly enhance control and management strategies of their power output. The paper introduces a new methodology for modelling and forecasting such very short-term fluctuations. The proposed...... consists in 1-step ahead forecasting exercise on time-series of wind generation with a time resolution of 10 minute. The quality of the introduced forecasting methodology and its interest for better understanding power fluctuations are finally discussed....... methodology is based on a Markov-switching autoregressive model with time-varying coefficients. An advantage of the method is that one can easily derive full predictive densities. The quality of this methodology is demonstrated from the test case of 2 large offshore wind farms in Denmark. The exercise...
The Astringency of the GP Algorithm for Forecasting Software Failure Data Series
Directory of Open Access Journals (Sweden)
Yong-qiang Zhang
2007-05-01
Full Text Available The forecasting of software failure data series by Genetic Programming (GP can be realized without any assumptions before modeling. This discovery has transformed traditional statistical modeling methods as well as improved consistency for model applicability. The individuals' different characteristics during the evolution of generations, which are randomly changeable, are treated as Markov random processes. This paper also proposes that a GP algorithm with "optimal individuals reserved strategy" is the best solution to this problem, and therefore the adaptive individuals finally will be evolved. This will allow practical applications in software reliability modeling analysis and forecasting for failure behaviors. Moreover it can verify the feasibility and availability of the GP algorithm, which is applied to software failure data series forecasting on a theoretical basis. The results show that the GP algorithm is the best solution for software failure behaviors in a variety of disciplines.
Linear and nonlinear dynamic systems in financial time series prediction
Directory of Open Access Journals (Sweden)
Salim Lahmiri
2012-10-01
Full Text Available Autoregressive moving average (ARMA process and dynamic neural networks namely the nonlinear autoregressive moving average with exogenous inputs (NARX are compared by evaluating their ability to predict financial time series; for instance the S&P500 returns. Two classes of ARMA are considered. The first one is the standard ARMA model which is a linear static system. The second one uses Kalman filter (KF to estimate and predict ARMA coefficients. This model is a linear dynamic system. The forecasting ability of each system is evaluated by means of mean absolute error (MAE and mean absolute deviation (MAD statistics. Simulation results indicate that the ARMA-KF system performs better than the standard ARMA alone. Thus, introducing dynamics into the ARMA process improves the forecasting accuracy. In addition, the ARMA-KF outperformed the NARX. This result may suggest that the linear component found in the S&P500 return series is more dominant than the nonlinear part. In sum, we conclude that introducing dynamics into the ARMA process provides an effective system for S&P500 time series prediction.
Advances in Antithetic Time Series Analysis : Separating Fact from Artifact
Directory of Open Access Journals (Sweden)
Dennis Ridley
2016-01-01
Full Text Available The problem of biased time series mathematical model parameter estimates is well known to be insurmountable. When used to predict future values by extrapolation, even a de minimis bias will eventually grow into a large bias, with misleading results. This paper elucidates how combining antithetic time series' solves this baffling problem of bias in the fitted and forecast values by dynamic bias cancellation. Instead of growing to infinity, the average error can converge to a constant. (original abstract
SHORT-TERM FORECASTING OF MORTGAGE LENDING
Directory of Open Access Journals (Sweden)
Irina V. Orlova
2013-01-01
Full Text Available The article considers the methodological and algorithmic problems arising in modeling and forecasting of time series of mortgage loans. Focuses on the processes of formation of the levels of time series of mortgage loans and the problem of choice and identification of models in the conditions of small samples. For forecasting options are selected and implemented a model of autoregressive and moving average, which allowed to obtain reliable forecasts.
A neuro-fuzzy computing technique for modeling hydrological time series
Nayak, P. C.; Sudheer, K. P.; Rangan, D. M.; Ramasastri, K. S.
2004-05-01
Intelligent computing tools such as artificial neural network (ANN) and fuzzy logic approaches are proven to be efficient when applied individually to a variety of problems. Recently there has been a growing interest in combining both these approaches, and as a result, neuro-fuzzy computing techniques have evolved. This approach has been tested and evaluated in the field of signal processing and related areas, but researchers have only begun evaluating the potential of this neuro-fuzzy hybrid approach in hydrologic modeling studies. This paper presents the application of an adaptive neuro fuzzy inference system (ANFIS) to hydrologic time series modeling, and is illustrated by an application to model the river flow of Baitarani River in Orissa state, India. An introduction to the ANFIS modeling approach is also presented. The advantage of the method is that it does not require the model structure to be known a priori, in contrast to most of the time series modeling techniques. The results showed that the ANFIS forecasted flow series preserves the statistical properties of the original flow series. The model showed good performance in terms of various statistical indices. The results are highly promising, and a comparative analysis suggests that the proposed modeling approach outperforms ANNs and other traditional time series models in terms of computational speed, forecast errors, efficiency, peak flow estimation etc. It was observed that the ANFIS model preserves the potential of the ANN approach fully, and eases the model building process.
Real-time data processing and inflow forecasting
International Nuclear Information System (INIS)
Olason, T.; Lafreniere, M.
1998-01-01
One of the key inputs into the short-term scheduling of hydroelectric generation is inflow forecasting which is needed for natural or unregulated inflows into various lakes, reservoirs and river sections. The forecast time step and time horizon are determined by the time step and the scheduling horizon. Acres International Ltd. has developed the Vista Decision Support System (DSS) in which the time step is one hour and the scheduling can be done up to two weeks into the future. This paper presents the basis of the operational flow-forecasting module of the Vista DSS software and its application to flow forecasting for 16 basins within Nova Scotia Power's hydroelectric system. Among the tasks performed by the software are collection and treatment of data (in real time) regarding meteorological forecasts, reviews and monitoring of hydro-meteorological data, updating of the state variables in the module, and the review and adjustment of sub-watershed forecasts
factor high order fuzzy time series with applications to temperature
African Journals Online (AJOL)
HOD
In this paper, a novel two – factor high – order fuzzy time series forecasting method based on .... to balance between local and global exploitations of the swarms. While, .... Although, there were a number of outliers but, the spread at the spot in ...
Knowing what to expect, forecasting monthly emergency department visits: A time-series analysis.
Bergs, Jochen; Heerinckx, Philipe; Verelst, Sandra
2014-04-01
To evaluate an automatic forecasting algorithm in order to predict the number of monthly emergency department (ED) visits one year ahead. We collected retrospective data of the number of monthly visiting patients for a 6-year period (2005-2011) from 4 Belgian Hospitals. We used an automated exponential smoothing approach to predict monthly visits during the year 2011 based on the first 5 years of the dataset. Several in- and post-sample forecasting accuracy measures were calculated. The automatic forecasting algorithm was able to predict monthly visits with a mean absolute percentage error ranging from 2.64% to 4.8%, indicating an accurate prediction. The mean absolute scaled error ranged from 0.53 to 0.68 indicating that, on average, the forecast was better compared with in-sample one-step forecast from the naïve method. The applied automated exponential smoothing approach provided useful predictions of the number of monthly visits a year in advance. Copyright © 2013 Elsevier Ltd. All rights reserved.
Origins of forecast skill of weather and climate events on verifiable time scales
CSIR Research Space (South Africa)
Landman, WA
2012-07-01
Full Text Available specific location between the predictor or the predictand and their respective canonical component time series (rj and sk) Barnett, T. P., and Preisendorfer, R. W. 1987: Origins and levels of monthly and seasonal forecast skill for United States air...
Time series analysis of gold production in Malaysia
Muda, Nora; Hoon, Lee Yuen
2012-05-01
Gold is a soft, malleable, bright yellow metallic element and unaffected by air or most reagents. It is highly valued as an asset or investment commodity and is extensively used in jewellery, industrial application, dentistry and medical applications. In Malaysia, gold mining is limited in several areas such as Pahang, Kelantan, Terengganu, Johor and Sarawak. The main purpose of this case study is to obtain a suitable model for the production of gold in Malaysia. The model can also be used to predict the data of Malaysia's gold production in the future. Box-Jenkins time series method was used to perform time series analysis with the following steps: identification, estimation, diagnostic checking and forecasting. In addition, the accuracy of prediction is tested using mean absolute percentage error (MAPE). From the analysis, the ARIMA (3,1,1) model was found to be the best fitted model with MAPE equals to 3.704%, indicating the prediction is very accurate. Hence, this model can be used for forecasting. This study is expected to help the private and public sectors to understand the gold production scenario and later plan the gold mining activities in Malaysia.
Time series analysis of monthly pulpwood use in the Northeast
James T. Bones
1980-01-01
Time series analysis was used to develop a model that depicts pulpwood use in the Northeast. The model is useful in forecasting future pulpwood requirements (short term) or monitoring pulpwood-use activity in relation to past use patterns. The model predicted a downturn in use during 1980.
Guarnaccia, Claudio; Quartieri, Joseph; Tepedino, Carmine
2017-06-01
One of the most hazardous physical polluting agents, considering their effects on human health, is acoustical noise. Airports are a strong source of acoustical noise, due to the airplanes turbines, to the aero-dynamical noise of transits, to the acceleration or the breaking during the take-off and landing phases of aircrafts, to the road traffic around the airport, etc.. The monitoring and the prediction of the acoustical level emitted by airports can be very useful to assess the impact on human health and activities. In the airports noise scenario, thanks to flights scheduling, the predominant sources may have a periodic behaviour. Thus, a Time Series Analysis approach can be adopted, considering that a general trend and a seasonal behaviour can be highlighted and used to build a predictive model. In this paper, two different approaches are adopted, thus two predictive models are constructed and tested. The first model is based on deterministic decomposition and is built composing the trend, that is the long term behaviour, the seasonality, that is the periodic component, and the random variations. The second model is based on seasonal autoregressive moving average, and it belongs to the stochastic class of models. The two different models are fitted on an acoustical level dataset collected close to the Nice (France) international airport. Results will be encouraging and will show good prediction performances of both the adopted strategies. A residual analysis is performed, in order to quantify the forecasting error features.
Big Data impacts on stochastic Forecast Models: Evidence from FX time series
Directory of Open Access Journals (Sweden)
Sebastian Dietz
2013-12-01
Full Text Available With the rise of the Big Data paradigm new tasks for prediction models appeared. In addition to the volume problem of such data sets nonlinearity becomes important, as the more detailed data sets contain also more comprehensive information, e.g. about non regular seasonal or cyclical movements as well as jumps in time series. This essay compares two nonlinear methods for predicting a high frequency time series, the USD/Euro exchange rate. The first method investigated is Autoregressive Neural Network Processes (ARNN, a neural network based nonlinear extension of classical autoregressive process models from time series analysis (see Dietz 2011. Its advantage is its simple but scalable time series process model architecture, which is able to include all kinds of nonlinearities based on the universal approximation theorem of Hornik, Stinchcombe and White 1989 and the extensions of Hornik 1993. However, restrictions related to the numeric estimation procedures limit the flexibility of the model. The alternative is a Support Vector Machine Model (SVM, Vapnik 1995. The two methods compared have different approaches of error minimization (Empirical error minimization at the ARNN vs. structural error minimization at the SVM. Our new finding is, that time series data classified as “Big Data” need new methods for prediction. Estimation and prediction was performed using the statistical programming language R. Besides prediction results we will also discuss the impact of Big Data on data preparation and model validation steps. Normal 0 21 false false false DE X-NONE X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Normale Tabelle"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-parent:""; mso-padding-alt:0cm 5.4pt 0cm 5.4pt; mso-para-margin:0cm; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:10.0pt; font-family:"Times New Roman","serif";}
Self-organising mixture autoregressive model for non-stationary time series modelling.
Ni, He; Yin, Hujun
2008-12-01
Modelling non-stationary time series has been a difficult task for both parametric and nonparametric methods. One promising solution is to combine the flexibility of nonparametric models with the simplicity of parametric models. In this paper, the self-organising mixture autoregressive (SOMAR) network is adopted as a such mixture model. It breaks time series into underlying segments and at the same time fits local linear regressive models to the clusters of segments. In such a way, a global non-stationary time series is represented by a dynamic set of local linear regressive models. Neural gas is used for a more flexible structure of the mixture model. Furthermore, a new similarity measure has been introduced in the self-organising network to better quantify the similarity of time series segments. The network can be used naturally in modelling and forecasting non-stationary time series. Experiments on artificial, benchmark time series (e.g. Mackey-Glass) and real-world data (e.g. numbers of sunspots and Forex rates) are presented and the results show that the proposed SOMAR network is effective and superior to other similar approaches.
Ike, P.; Voogd, Henk; Voogd, Henk; Linden, Gerard
2004-01-01
This chapter begins with a discussion of qualitative forecasting by describing a number of methods that depend on judgements made by stakeholders, experts or other interested parties to arrive at forecasts. Two qualitative approaches are illuminated, the Delphi and scenario methods respectively. Quantitative forecasting is illustrated with a brief overview of time series methods. Both qualitative and quantitative methods are illustrated by an example. The role and relative importance of forec...
International Nuclear Information System (INIS)
Chou, Jui-Sheng; Ngo, Ngoc-Tri
2016-01-01
Highlights: • This study develops a novel time-series sliding window forecast system. • The system integrates metaheuristics, machine learning and time-series models. • Site experiment of smart grid infrastructure is installed to retrieve real-time data. • The proposed system accurately predicts energy consumption in residential buildings. • The forecasting system can help users minimize their electricity usage. - Abstract: Smart grids are a promising solution to the rapidly growing power demand because they can considerably increase building energy efficiency. This study developed a novel time-series sliding window metaheuristic optimization-based machine learning system for predicting real-time building energy consumption data collected by a smart grid. The proposed system integrates a seasonal autoregressive integrated moving average (SARIMA) model and metaheuristic firefly algorithm-based least squares support vector regression (MetaFA-LSSVR) model. Specifically, the proposed system fits the SARIMA model to linear data components in the first stage, and the MetaFA-LSSVR model captures nonlinear data components in the second stage. Real-time data retrieved from an experimental smart grid installed in a building were used to evaluate the efficacy and effectiveness of the proposed system. A k-week sliding window approach is proposed for employing historical data as input for the novel time-series forecasting system. The prediction system yielded high and reliable accuracy rates in 1-day-ahead predictions of building energy consumption, with a total error rate of 1.181% and mean absolute error of 0.026 kW h. Notably, the system demonstrates an improved accuracy rate in the range of 36.8–113.2% relative to those of the linear forecasting model (i.e., SARIMA) and nonlinear forecasting models (i.e., LSSVR and MetaFA-LSSVR). Therefore, end users can further apply the forecasted information to enhance efficiency of energy usage in their buildings, especially
Time series analysis of barometric pressure data
International Nuclear Information System (INIS)
La Rocca, Paola; Riggi, Francesco; Riggi, Daniele
2010-01-01
Time series of atmospheric pressure data, collected over a period of several years, were analysed to provide undergraduate students with educational examples of application of simple statistical methods of analysis. In addition to basic methods for the analysis of periodicities, a comparison of two forecast models, one based on autoregression algorithms, and the other making use of an artificial neural network, was made. Results show that the application of artificial neural networks may give slightly better results compared to traditional methods.
With string model to time series forecasting
Pinčák, Richard; Bartoš, Erik
2015-10-01
Overwhelming majority of econometric models applied on a long term basis in the financial forex market do not work sufficiently well. The reason is that transaction costs and arbitrage opportunity are not included, as this does not simulate the real financial markets. Analyses are not conducted on the non equidistant date but rather on the aggregate date, which is also not a real financial case. In this paper, we would like to show a new way how to analyze and, moreover, forecast financial market. We utilize the projections of the real exchange rate dynamics onto the string-like topology in the OANDA market. The latter approach allows us to build the stable prediction models in trading in the financial forex market. The real application of the multi-string structures is provided to demonstrate our ideas for the solution of the problem of the robust portfolio selection. The comparison with the trend following strategies was performed, the stability of the algorithm on the transaction costs for long trade periods was confirmed.
Time series forecasting using ERNN and QR based on Bayesian model averaging
Pwasong, Augustine; Sathasivam, Saratha
2017-08-01
The Bayesian model averaging technique is a multi-model combination technique. The technique was employed to amalgamate the Elman recurrent neural network (ERNN) technique with the quadratic regression (QR) technique. The amalgamation produced a hybrid technique known as the hybrid ERNN-QR technique. The potentials of forecasting with the hybrid technique are compared with the forecasting capabilities of individual techniques of ERNN and QR. The outcome revealed that the hybrid technique is superior to the individual techniques in the mean square error sense.
Aulia, D.; Ayu, S. F.; Matondang, A.
2018-01-01
Malaria is the most contagious global concern. As a public health problem with outbreaks, affect the quality of life and economy, also could lead to death. Therefore, this research is to forecast malaria cases with climatic factors as predictors in Mandailing Natal Regency. The total number of positive malaria cases on January 2008 to December 2016 were taken from health department of Mandailing Natal Regency. Climates data such as rainfall, humidity, and temperature were taken from Center of Statistic Department of Mandailing Natal Regency. E-views ver. 9 is used to analyze this study. Autoregressive integrated average, ARIMA (0,1,1) (1,0,0)12 is the best model to explain the 67,2% variability data in time series study. Rainfall (P value = 0.0005), temperature (P value = 0,0029) and humidity (P value = 0.0001) are significant predictors for malaria transmission. Seasonal adjusted factor (SAF) in November and March shows peak for malaria cases.
A Feature Fusion Based Forecasting Model for Financial Time Series
Guo, Zhiqiang; Wang, Huaiqing; Liu, Quan; Yang, Jie
2014-01-01
Predicting the stock market has become an increasingly interesting research area for both researchers and investors, and many prediction models have been proposed. In these models, feature selection techniques are used to pre-process the raw data and remove noise. In this paper, a prediction model is constructed to forecast stock market behavior with the aid of independent component analysis, canonical correlation analysis, and a support vector machine. First, two types of features are extracted from the historical closing prices and 39 technical variables obtained by independent component analysis. Second, a canonical correlation analysis method is utilized to combine the two types of features and extract intrinsic features to improve the performance of the prediction model. Finally, a support vector machine is applied to forecast the next day's closing price. The proposed model is applied to the Shanghai stock market index and the Dow Jones index, and experimental results show that the proposed model performs better in the area of prediction than other two similar models. PMID:24971455
Directory of Open Access Journals (Sweden)
Chi Zhang
2013-03-01
Full Text Available Purpose: The purpose of this paper is to develop a combined model composed of grey-forecast model and Logistic-growth-curve model to improve the accuracy of forecast model of cargo throughput for the port. The authors also use the existing data of a current port to verify the validity of the combined model.Design/methodology/approach: A literature review is undertaken to find the appropriate forecast model of cargo throughput for the port. Through researching the related forecast model, the authors put together the individual models which are significant to study further. Finally, the authors combine two individual models (grey-forecast model and Logistic-growth-curve model into one combined model to forecast the port cargo throughput, and use the model to a physical port in China to testify the validity of the model.Findings: Test by the perceptional data of cargo throughput in the physical port, the results show that the combined model can obtain relatively higher forecast accuracy when it is not easy to find more information. Furthermore, the forecast made by the combined model are more accurate than any of the individual ones.Research limitations/implications: The study provided a new combined forecast model of cargo throughput with a relatively less information to improve the accuracy rate of the forecast. The limitation of the model is that it requires the cargo throughput of the port have an S-shaped change trend.Practical implications: This model is not limited by external conditions such as geographical, cultural. This model predicted the port cargo throughput of one real port in China in 2015, which provided some instructive guidance for the port development.Originality/value: This is the one of the study to improve the accuracy rate of the cargo throughput forecast with little information.
Effect of Streamflow Forecast Uncertainty on Real-Time Reservoir Operation
Zhao, T.; Cai, X.; Yang, D.
2010-12-01
Various hydrological forecast products have been applied to real-time reservoir operation, including deterministic streamflow forecast (DSF), DSF-based probabilistic streamflow forecast (DPSF), and ensemble streamflow forecast (ESF), which represent forecast uncertainty in the form of deterministic forecast error, deterministic forecast error-based uncertainty distribution, and ensemble forecast errors, respectively. Compared to previous studies that treat these forecast products as ad hoc inputs for reservoir operation models, this paper attempts to model the uncertainties involved in the various forecast products and explores their effect on real-time reservoir operation decisions. In hydrology, there are various indices reflecting the magnitude of streamflow forecast uncertainty; meanwhile, few models illustrate the forecast uncertainty evolution process. This research introduces Martingale Model of Forecast Evolution (MMFE) from supply chain management and justifies its assumptions for quantifying the evolution of uncertainty in streamflow forecast as time progresses. Based on MMFE, this research simulates the evolution of forecast uncertainty in DSF, DPSF, and ESF, and applies the reservoir operation models (dynamic programming, DP; stochastic dynamic programming, SDP; and standard operation policy, SOP) to assess the effect of different forms of forecast uncertainty on real-time reservoir operation. Through a hypothetical single-objective real-time reservoir operation model, the results illustrate that forecast uncertainty exerts significant effects. Reservoir operation efficiency, as measured by a utility function, decreases as the forecast uncertainty increases. Meanwhile, these effects also depend on the type of forecast product being used. In general, the utility of reservoir operation with ESF is nearly as high as the utility obtained with a perfect forecast; the utilities of DSF and DPSF are similar to each other but not as efficient as ESF. Moreover
Energy Technology Data Exchange (ETDEWEB)
Zhen-Tang Liu; En-Lai Zhao; En-Yuan Wang; Jing Wang [China University of Mining and Technology, Xuzhou (China). School of Safety Engineering
2009-02-15
Based on chaos theory, the chaotic characteristics of electromagnetic radiation time series of coal or rock under different loads was studied. The results show that the correlation of electromagnetic radiation time series of small-scale coal or rock and coal mine converges to a stable saturation value, which shows that these electromagnetic radiation time series have chaos characteristics. When there is danger of coal seam burst, the value of the saturation correlation dimension D{sub 2} of the electromagnetic radiation time series is bigger and it changes greatly; when there is no danger, its value is smaller and changes smoothly. The change of saturation correlation of electromagnetic radiation time series can be used to forecast coal or rock dynamic disasters. 11 refs., 4 figs.
Self-affinity in the dengue fever time series
Azevedo, S. M.; Saba, H.; Miranda, J. G. V.; Filho, A. S. Nascimento; Moret, M. A.
2016-06-01
Dengue is a complex public health problem that is common in tropical and subtropical regions. This disease has risen substantially in the last three decades, and the physical symptoms depict the self-affine behavior of the occurrences of reported dengue cases in Bahia, Brazil. This study uses detrended fluctuation analysis (DFA) to verify the scale behavior in a time series of dengue cases and to evaluate the long-range correlations that are characterized by the power law α exponent for different cities in Bahia, Brazil. The scaling exponent (α) presents different long-range correlations, i.e. uncorrelated, anti-persistent, persistent and diffusive behaviors. The long-range correlations highlight the complex behavior of the time series of this disease. The findings show that there are two distinct types of scale behavior. In the first behavior, the time series presents a persistent α exponent for a one-month period. For large periods, the time series signal approaches subdiffusive behavior. The hypothesis of the long-range correlations in the time series of the occurrences of reported dengue cases was validated. The observed self-affinity is useful as a forecasting tool for future periods through extrapolation of the α exponent behavior. This complex system has a higher predictability in a relatively short time (approximately one month), and it suggests a new tool in epidemiological control strategies. However, predictions for large periods using DFA are hidden by the subdiffusive behavior.
Single-Index Additive Vector Autoregressive Time Series Models
LI, YEHUA
2009-09-01
We study a new class of nonlinear autoregressive models for vector time series, where the current vector depends on single-indexes defined on the past lags and the effects of different lags have an additive form. A sufficient condition is provided for stationarity of such models. We also study estimation of the proposed model using P-splines, hypothesis testing, asymptotics, selection of the order of the autoregression and of the smoothing parameters and nonlinear forecasting. We perform simulation experiments to evaluate our model in various settings. We illustrate our methodology on a climate data set and show that our model provides more accurate yearly forecasts of the El Niño phenomenon, the unusual warming of water in the Pacific Ocean. © 2009 Board of the Foundation of the Scandinavian Journal of Statistics.
Daily rainfall forecasting for one year in a single run using Singular Spectrum Analysis
Unnikrishnan, Poornima; Jothiprakash, V.
2018-06-01
Effective modelling and prediction of smaller time step rainfall is reported to be very difficult owing to its highly erratic nature. Accurate forecast of daily rainfall for longer duration (multi time step) may be exceptionally helpful in the efficient planning and management of water resources systems. Identification of inherent patterns in a rainfall time series is also important for an effective water resources planning and management system. In the present study, Singular Spectrum Analysis (SSA) is utilized to forecast the daily rainfall time series pertaining to Koyna watershed in Maharashtra, India, for 365 days after extracting various components of the rainfall time series such as trend, periodic component, noise and cyclic component. In order to forecast the time series for longer time step (365 days-one window length), the signal and noise components of the time series are forecasted separately and then added together. The results of the study show that the method of SSA could extract the various components of the time series effectively and could also forecast the daily rainfall time series for longer duration such as one year in a single run with reasonable accuracy.
Kadri, Farid
2018-02-07
Efficient management of patient demands in emergency departments (EDs) has recently received increasing attention by most healthcare administrations. Forecasting ED demands greatly helps ED\\'s managers to make suitable decisions by optimally allocating the available limited resources to efficiently handle patient attendances. Furthermore, it permits pre-emptive action(s) to mitigate and/or prevent overcrowding situations and to enhance the quality of care. In this work, we present a statistical approach based on a vector autoregressive moving average (VARMA) model for a short term forecasting of daily attendances at an ED. The VARMA model has been validated using an experimental data from the paediatric emergency department (PED) at Lille regional hospital centre, France. The results obtained indicate the effectiveness of the proposed approach in forecasting patient demands.
Short-Term Wave Forecasting for Real-Time Control of Wave Energy Converters
Fusco, Francesco; Ringwood, John
2010-01-01
Real-time control of wave energy converters requires knowledge of future incident wave elevation in order to approach optimal efficiency of wave energy extraction. We present an approach where the wave elevation is treated as a time series and it is predicted only from its past history. A comparison of a range of forecasting methodologies on real wave observations from two different locations shows how the relatively simple linear autoregressive model, which implicitly models the cyclical beh...
Multiresolution forecasting for futures trading using wavelet decompositions.
Zhang, B L; Coggins, R; Jabri, M A; Dersch, D; Flower, B
2001-01-01
We investigate the effectiveness of a financial time-series forecasting strategy which exploits the multiresolution property of the wavelet transform. A financial series is decomposed into an over complete, shift invariant scale-related representation. In transform space, each individual wavelet series is modeled by a separate multilayer perceptron (MLP). We apply the Bayesian method of automatic relevance determination to choose short past windows (short-term history) for the inputs to the MLPs at lower scales and long past windows (long-term history) at higher scales. To form the overall forecast, the individual forecasts are then recombined by the linear reconstruction property of the inverse transform with the chosen autocorrelation shell representation, or by another perceptron which learns the weight of each scale in the prediction of the original time series. The forecast results are then passed to a money management system to generate trades.
Can we use Earth Observations to improve monthly water level forecasts?
Slater, L. J.; Villarini, G.
2017-12-01
Dynamical-statistical hydrologic forecasting approaches benefit from different strengths in comparison with traditional hydrologic forecasting systems: they are computationally efficient, can integrate and `learn' from a broad selection of input data (e.g., General Circulation Model (GCM) forecasts, Earth Observation time series, teleconnection patterns), and can take advantage of recent progress in machine learning (e.g. multi-model blending, post-processing and ensembling techniques). Recent efforts to develop a dynamical-statistical ensemble approach for forecasting seasonal streamflow using both GCM forecasts and changing land cover have shown promising results over the U.S. Midwest. Here, we use climate forecasts from several GCMs of the North American Multi Model Ensemble (NMME) alongside 15-minute stage time series from the National River Flow Archive (NRFA) and land cover classes extracted from the European Space Agency's Climate Change Initiative 300 m annual Global Land Cover time series. With these data, we conduct systematic long-range probabilistic forecasting of monthly water levels in UK catchments over timescales ranging from one to twelve months ahead. We evaluate the improvement in model fit and model forecasting skill that comes from using land cover classes as predictors in the models. This work opens up new possibilities for combining Earth Observation time series with GCM forecasts to predict a variety of hazards from space using data science techniques.
Patra, S. R.
2017-12-01
Evapotranspiration (ET0) influences water resources and it is considered as a vital process in aridic hydrologic frameworks. It is one of the most important measure in finding the drought condition. Therefore, time series forecasting of evapotranspiration is very important in order to help the decision makers and water system mangers build up proper systems to sustain and manage water resources. Time series considers that -history repeats itself, hence by analysing the past values, better choices, or forecasts, can be carried out for the future. Ten years of ET0 data was used as a part of this study to make sure a satisfactory forecast of monthly values. In this study, three models: (ARIMA) mathematical model, artificial neural network model, support vector machine model are presented. These three models are used for forecasting monthly reference crop evapotranspiration based on ten years of past historical records (1991-2001) of measured evaporation at Ganjam region, Odisha, India without considering the climate data. The developed models will allow water resource managers to predict up to 12 months, making these predictions very useful to optimize the resources needed for effective water resources management. In this study multistep-ahead prediction is performed which is more complex and troublesome than onestep ahead. Our investigation proposed that nonlinear relationships may exist among the monthly indices, so that the ARIMA model might not be able to effectively extract the full relationship hidden in the historical data. Support vector machines are potentially helpful time series forecasting strategies on account of their strong nonlinear mapping capability and resistance to complexity in forecasting data. SVMs have great learning capability in time series modelling compared to ANN. For instance, the SVMs execute the structural risk minimization principle, which allows in better generalization as compared to neural networks that use the empirical risk
Uncertainties in Forecasting Streamflow using Entropy Theory
Cui, H.; Singh, V. P.
2017-12-01
Streamflow forecasting is essential in river restoration, reservoir operation, power generation, irrigation, navigation, and water management. However, there is always uncertainties accompanied in forecast, which may affect the forecasting results and lead to large variations. Therefore, uncertainties must be considered and be assessed properly when forecasting streamflow for water management. The aim of our work is to quantify the uncertainties involved in forecasting streamflow and provide reliable streamflow forecast. Despite that streamflow time series are stochastic, they exhibit seasonal and periodic patterns. Therefore, streamflow forecasting entails modeling seasonality, periodicity, and its correlation structure, and assessing uncertainties. This study applies entropy theory to forecast streamflow and measure uncertainties during the forecasting process. To apply entropy theory for streamflow forecasting, spectral analysis is combined to time series analysis, as spectral analysis can be employed to characterize patterns of streamflow variation and identify the periodicity of streamflow. That is, it permits to extract significant information for understanding the streamflow process and prediction thereof. Application of entropy theory for streamflow forecasting involves determination of spectral density, determination of parameters, and extension of autocorrelation function. The uncertainties brought by precipitation input, forecasting model and forecasted results are measured separately using entropy. With information theory, how these uncertainties transported and aggregated during these processes will be described.
Forecasting with Dynamic Regression Models
Pankratz, Alan
2012-01-01
One of the most widely used tools in statistical forecasting, single equation regression models is examined here. A companion to the author's earlier work, Forecasting with Univariate Box-Jenkins Models: Concepts and Cases, the present text pulls together recent time series ideas and gives special attention to possible intertemporal patterns, distributed lag responses of output to input series and the auto correlation patterns of regression disturbance. It also includes six case studies.
International Nuclear Information System (INIS)
Long, L.L.
1976-01-01
Fourier series and statistical communication theory techniques are utilized in the estimation of river water temperature increases caused by external thermal inputs. An example estimate assuming a constant thermal input is demonstrated. A regression fit of the Fourier series approximation of temperature is then used to forecast daily average water temperatures. Also, a 60-day prediction of daily average water temperature is made with the aid of the Fourier regression fit by using significant Fourier components
Non-parametric characterization of long-term rainfall time series
Tiwari, Harinarayan; Pandey, Brij Kishor
2018-03-01
The statistical study of rainfall time series is one of the approaches for efficient hydrological system design. Identifying, and characterizing long-term rainfall time series could aid in improving hydrological systems forecasting. In the present study, eventual statistics was applied for the long-term (1851-2006) rainfall time series under seven meteorological regions of India. Linear trend analysis was carried out using Mann-Kendall test for the observed rainfall series. The observed trend using the above-mentioned approach has been ascertained using the innovative trend analysis method. Innovative trend analysis has been found to be a strong tool to detect the general trend of rainfall time series. Sequential Mann-Kendall test has also been carried out to examine nonlinear trends of the series. The partial sum of cumulative deviation test is also found to be suitable to detect the nonlinear trend. Innovative trend analysis, sequential Mann-Kendall test and partial cumulative deviation test have potential to detect the general as well as nonlinear trend for the rainfall time series. Annual rainfall analysis suggests that the maximum changes in mean rainfall is 11.53% for West Peninsular India, whereas the maximum fall in mean rainfall is 7.8% for the North Mountainous Indian region. The innovative trend analysis method is also capable of finding the number of change point available in the time series. Additionally, we have performed von Neumann ratio test and cumulative deviation test to estimate the departure from homogeneity. Singular spectrum analysis has been applied in this study to evaluate the order of departure from homogeneity in the rainfall time series. Monsoon season (JS) of North Mountainous India and West Peninsular India zones has higher departure from homogeneity and singular spectrum analysis shows the results to be in coherence with the same.
Grey Forecast Rainfall with Flow Updating Algorithm for Real-Time Flood Forecasting
Directory of Open Access Journals (Sweden)
Jui-Yi Ho
2015-04-01
Full Text Available The dynamic relationship between watershed characteristics and rainfall-runoff has been widely studied in recent decades. Since watershed rainfall-runoff is a non-stationary process, most deterministic flood forecasting approaches are ineffective without the assistance of adaptive algorithms. The purpose of this paper is to propose an effective flow forecasting system that integrates a rainfall forecasting model, watershed runoff model, and real-time updating algorithm. This study adopted a grey rainfall forecasting technique, based on existing hourly rainfall data. A geomorphology-based runoff model can be used for simulating impacts of the changing geo-climatic conditions on the hydrologic response of unsteady and non-linear watershed system, and flow updating algorithm were combined to estimate watershed runoff according to measured flow data. The proposed flood forecasting system was applied to three watersheds; one in the United States and two in Northern Taiwan. Four sets of rainfall-runoff simulations were performed to test the accuracy of the proposed flow forecasting technique. The results indicated that the forecast and observed hydrographs are in good agreement for all three watersheds. The proposed flow forecasting system could assist authorities in minimizing loss of life and property during flood events.
Lohani, A. K.; Kumar, Rakesh; Singh, R. D.
2012-06-01
SummaryTime series modeling is necessary for the planning and management of reservoirs. More recently, the soft computing techniques have been used in hydrological modeling and forecasting. In this study, the potential of artificial neural networks and neuro-fuzzy system in monthly reservoir inflow forecasting are examined by developing and comparing monthly reservoir inflow prediction models, based on autoregressive (AR), artificial neural networks (ANNs) and adaptive neural-based fuzzy inference system (ANFIS). To take care the effect of monthly periodicity in the flow data, cyclic terms are also included in the ANN and ANFIS models. Working with time series flow data of the Sutlej River at Bhakra Dam, India, several ANN and adaptive neuro-fuzzy models are trained with different input vectors. To evaluate the performance of the selected ANN and adaptive neural fuzzy inference system (ANFIS) models, comparison is made with the autoregressive (AR) models. The ANFIS model trained with the input data vector including previous inflows and cyclic terms of monthly periodicity has shown a significant improvement in the forecast accuracy in comparison with the ANFIS models trained with the input vectors considering only previous inflows. In all cases ANFIS gives more accurate forecast than the AR and ANN models. The proposed ANFIS model coupled with the cyclic terms is shown to provide better representation of the monthly inflow forecasting for planning and operation of reservoir.
Time series analysis of the behavior of brazilian natural rubber
Directory of Open Access Journals (Sweden)
Antônio Donizette de Oliveira
2009-03-01
Full Text Available The natural rubber is a non-wood product obtained of the coagulation of some lattices of forest species, being Hevea brasiliensis the main one. Native from the Amazon Region, this species was already known by the Indians before the discovery of America. The natural rubber became a product globally valued due to its multiple applications in the economy, being its almost perfect substitute the synthetic rubber derived from the petroleum. Similarly to what happens with other countless products the forecast of future prices of the natural rubber has been object of many studies. The use of models of forecast of univariate timeseries stands out as the more accurate and useful to reduce the uncertainty in the economic decision making process. This studyanalyzed the historical series of prices of the Brazilian natural rubber (R$/kg, in the Jan/99 - Jun/2006 period, in order tocharacterize the rubber price behavior in the domestic market; estimated a model for the time series of monthly natural rubberprices; and foresaw the domestic prices of the natural rubber, in the Jul/2006 - Jun/2007 period, based on the estimated models.The studied models were the ones belonging to the ARIMA family. The main results were: the domestic market of the natural rubberis expanding due to the growth of the world economy; among the adjusted models, the ARIMA (1,1,1 model provided the bestadjustment of the time series of prices of the natural rubber (R$/kg; the prognosis accomplished for the series supplied statistically adequate fittings.
Forecasting daily patient volumes in the emergency department.
Jones, Spencer S; Thomas, Alun; Evans, R Scott; Welch, Shari J; Haug, Peter J; Snow, Gregory L
2008-02-01
Shifts in the supply of and demand for emergency department (ED) resources make the efficient allocation of ED resources increasingly important. Forecasting is a vital activity that guides decision-making in many areas of economic, industrial, and scientific planning, but has gained little traction in the health care industry. There are few studies that explore the use of forecasting methods to predict patient volumes in the ED. The goals of this study are to explore and evaluate the use of several statistical forecasting methods to predict daily ED patient volumes at three diverse hospital EDs and to compare the accuracy of these methods to the accuracy of a previously proposed forecasting method. Daily patient arrivals at three hospital EDs were collected for the period January 1, 2005, through March 31, 2007. The authors evaluated the use of seasonal autoregressive integrated moving average, time series regression, exponential smoothing, and artificial neural network models to forecast daily patient volumes at each facility. Forecasts were made for horizons ranging from 1 to 30 days in advance. The forecast accuracy achieved by the various forecasting methods was compared to the forecast accuracy achieved when using a benchmark forecasting method already available in the emergency medicine literature. All time series methods considered in this analysis provided improved in-sample model goodness of fit. However, post-sample analysis revealed that time series regression models that augment linear regression models by accounting for serial autocorrelation offered only small improvements in terms of post-sample forecast accuracy, relative to multiple linear regression models, while seasonal autoregressive integrated moving average, exponential smoothing, and artificial neural network forecasting models did not provide consistently accurate forecasts of daily ED volumes. This study confirms the widely held belief that daily demand for ED services is characterized by
A nonparametric approach to forecasting realized volatility
Adam Clements; Ralf Becker
2009-01-01
A well developed literature exists in relation to modeling and forecasting asset return volatility. Much of this relate to the development of time series models of volatility. This paper proposes an alternative method for forecasting volatility that does not involve such a model. Under this approach a forecast is a weighted average of historical volatility. The greatest weight is given to periods that exhibit the most similar market conditions to the time at which the forecast is being formed...
Time series behaviour of the number of Air Asia passengers: A distributional approach
Asrah, Norhaidah Mohd; Djauhari, Maman Abdurachman
2013-09-01
The common practice to time series analysis is by fitting a model and then further analysis is conducted on the residuals. However, if we know the distributional behavior of time series, the analyses in model identification, parameter estimation, and model checking are more straightforward. In this paper, we show that the number of Air Asia passengers can be represented as a geometric Brownian motion process. Therefore, instead of using the standard approach in model fitting, we use an appropriate transformation to come up with a stationary, normally distributed and even independent time series. An example in forecasting the number of Air Asia passengers will be given to illustrate the advantages of the method.
New prediction of chaotic time series based on local Lyapunov exponent
International Nuclear Information System (INIS)
Zhang Yong
2013-01-01
A new method of predicting chaotic time series is presented based on a local Lyapunov exponent, by quantitatively measuring the exponential rate of separation or attraction of two infinitely close trajectories in state space. After reconstructing state space from one-dimensional chaotic time series, neighboring multiple-state vectors of the predicting point are selected to deduce the prediction formula by using the definition of the local Lyapunov exponent. Numerical simulations are carried out to test its effectiveness and verify its higher precision over two older methods. The effects of the number of referential state vectors and added noise on forecasting accuracy are also studied numerically. (general)
Empirical testing of forecast update procedure forseasonal products
DEFF Research Database (Denmark)
Wong, Chee Yew; Johansen, John
2008-01-01
Updating of forecasts is essential for successful collaborative forecasting, especially for seasonal products. This paper discusses the results of a theoretical simulation and an empirical test of a proposed time-series forecast updating procedure. It involves a two-stage longitudinal case study...... of a toy supply chain. The theoretical simulation involves historical weekly consumer demand data for 122 toy products. The empirical test is then carried out in real-time with 291 toy products. The results show that the proposed forecast updating procedure: 1) reduced forecast errors of the annual...... provided less forecast accuracy improvement and it needed a longer time to achieve relatively acceptable forecast uncertainty....
Beat the Instructor: An Introductory Forecasting Game
Snider, Brent R.; Eliasson, Janice B.
2013-01-01
This teaching brief describes a 30-minute game where student groups compete in-class in an introductory time-series forecasting exercise. The students are challenged to "beat the instructor" who competes using forecasting techniques that will be subsequently taught. All forecasts are graphed prior to revealing the randomly generated…
International Nuclear Information System (INIS)
Wang, Yamin; Wu, Lei
2016-01-01
This paper presents a comprehensive analysis on practical challenges of empirical mode decomposition (EMD) based algorithms on wind speed and solar irradiation forecasts that have been largely neglected in literature, and proposes an alternative approach to mitigate such challenges. Specifically, the challenges are: (1) Decomposed sub-series are very sensitive to the original time series data. That is, sub-series of the new time series, consisting of the original one plus a limit number of new data samples, may significantly differ from those used in training forecasting models. In turn, forecasting models established by original sub-series may not be suitable for newly decomposed sub-series and have to be trained more frequently; and (2) Key environmental factors usually play a critical role in non-decomposition based methods for forecasting wind speed and solar irradiation. However, it is difficult to incorporate such critical environmental factors into forecasting models of individual decomposed sub-series, because the correlation between the original data and environmental factors is lost after decomposition. Numerical case studies on wind speed and solar irradiation forecasting show that the performance of existing EMD-based forecasting methods could be worse than the non-decomposition based forecasting model, and are not effective in practical cases. Finally, the approximated forecasting model based on EMD is proposed to mitigate the challenges and achieve better forecasting results than existing EMD-based forecasting algorithms and the non-decomposition based forecasting models on practical wind speed and solar irradiation forecasting cases. - Highlights: • Two challenges of existing EMD-based forecasting methods are discussed. • Significant changes of sub-series in each step of the rolling forecast procedure. • Difficulties in incorporating environmental factors into sub-series forecasting models. • The approximated forecasting method is proposed to
Comparative study of four time series methods in forecasting typhoid fever incidence in China.
Zhang, Xingyu; Liu, Yuanyuan; Yang, Min; Zhang, Tao; Young, Alistair A; Li, Xiaosong
2013-01-01
Accurate incidence forecasting of infectious disease is critical for early prevention and for better government strategic planning. In this paper, we present a comprehensive study of different forecasting methods based on the monthly incidence of typhoid fever. The seasonal autoregressive integrated moving average (SARIMA) model and three different models inspired by neural networks, namely, back propagation neural networks (BPNN), radial basis function neural networks (RBFNN), and Elman recurrent neural networks (ERNN) were compared. The differences as well as the advantages and disadvantages, among the SARIMA model and the neural networks were summarized and discussed. The data obtained for 2005 to 2009 and for 2010 from the Chinese Center for Disease Control and Prevention were used as modeling and forecasting samples, respectively. The performances were evaluated based on three metrics: mean absolute error (MAE), mean absolute percentage error (MAPE), and mean square error (MSE). The results showed that RBFNN obtained the smallest MAE, MAPE and MSE in both the modeling and forecasting processes. The performances of the four models ranked in descending order were: RBFNN, ERNN, BPNN and the SARIMA model.
Comparative study of four time series methods in forecasting typhoid fever incidence in China.
Directory of Open Access Journals (Sweden)
Xingyu Zhang
Full Text Available Accurate incidence forecasting of infectious disease is critical for early prevention and for better government strategic planning. In this paper, we present a comprehensive study of different forecasting methods based on the monthly incidence of typhoid fever. The seasonal autoregressive integrated moving average (SARIMA model and three different models inspired by neural networks, namely, back propagation neural networks (BPNN, radial basis function neural networks (RBFNN, and Elman recurrent neural networks (ERNN were compared. The differences as well as the advantages and disadvantages, among the SARIMA model and the neural networks were summarized and discussed. The data obtained for 2005 to 2009 and for 2010 from the Chinese Center for Disease Control and Prevention were used as modeling and forecasting samples, respectively. The performances were evaluated based on three metrics: mean absolute error (MAE, mean absolute percentage error (MAPE, and mean square error (MSE. The results showed that RBFNN obtained the smallest MAE, MAPE and MSE in both the modeling and forecasting processes. The performances of the four models ranked in descending order were: RBFNN, ERNN, BPNN and the SARIMA model.
Detecting and characterising ramp events in wind power time series
International Nuclear Information System (INIS)
Gallego, Cristóbal; Cuerva, Álvaro; Costa, Alexandre
2014-01-01
In order to implement accurate models for wind power ramp forecasting, ramps need to be previously characterised. This issue has been typically addressed by performing binary ramp/non-ramp classifications based on ad-hoc assessed thresholds. However, recent works question this approach. This paper presents the ramp function, an innovative wavelet- based tool which detects and characterises ramp events in wind power time series. The underlying idea is to assess a continuous index related to the ramp intensity at each time step, which is obtained by considering large power output gradients evaluated under different time scales (up to typical ramp durations). The ramp function overcomes some of the drawbacks shown by the aforementioned binary classification and permits forecasters to easily reveal specific features of the ramp behaviour observed at a wind farm. As an example, the daily profile of the ramp-up and ramp-down intensities are obtained for the case of a wind farm located in Spain
Setiawan, Suhartono, Ahmad, Imam Safawi; Rahmawati, Noorgam Ika
2015-12-01
Bank Indonesia (BI) as the central bank of Republic Indonesiahas a single overarching objective to establish and maintain rupiah stability. This objective could be achieved by monitoring traffic of inflow and outflow money currency. Inflow and outflow are related to stock and distribution of money currency around Indonesia territory. It will effect of economic activities. Economic activities of Indonesia,as one of Moslem country, absolutely related to Islamic Calendar (lunar calendar), that different with Gregorian calendar. This research aims to forecast the inflow and outflow money currency of Representative Office (RO) of BI Semarang Central Java region. The results of the analysis shows that the characteristics of inflow and outflow money currency influenced by the effects of the calendar variations, that is the day of Eid al-Fitr (moslem holyday) as well as seasonal patterns. In addition, the period of a certain week during Eid al-Fitr also affect the increase of inflow and outflow money currency. The best model based on the value of the smallestRoot Mean Square Error (RMSE) for inflow data is ARIMA model. While the best model for predicting the outflow data in RO of BI Semarang is ARIMAX model or Time Series Regression, because both of them have the same model. The results forecast in a period of 2015 shows an increase of inflow money currency happened in August, while the increase in outflow money currency happened in July.
Long-range dependence and sea level forecasting
Ercan, Ali; Abbasov, Rovshan K
2013-01-01
This study shows that the Caspian Sea level time series possess long range dependence even after removing linear trends, based on analyses of the Hurst statistic, the sample autocorrelation functions, and the periodogram of the series. Forecasting performance of ARMA, ARIMA, ARFIMA and Trend Line-ARFIMA (TL-ARFIMA) combination models are investigated. The forecast confidence bands and the forecast updating methodology, provided for ARIMA models in the literature, are modified for the ARFIMA models. Sample autocorrelation functions are utilized to estimate the differencing lengths of the ARFIMA
Wang, Wen-chuan; Chau, Kwok-wing; Qiu, Lin; Chen, Yang-bo
2015-05-01
Hydrological time series forecasting is one of the most important applications in modern hydrology, especially for the effective reservoir management. In this research, an artificial neural network (ANN) model coupled with the ensemble empirical mode decomposition (EEMD) is presented for forecasting medium and long-term runoff time series. First, the original runoff time series is decomposed into a finite and often small number of intrinsic mode functions (IMFs) and a residual series using EEMD technique for attaining deeper insight into the data characteristics. Then all IMF components and residue are predicted, respectively, through appropriate ANN models. Finally, the forecasted results of the modeled IMFs and residual series are summed to formulate an ensemble forecast for the original annual runoff series. Two annual reservoir runoff time series from Biuliuhe and Mopanshan in China, are investigated using the developed model based on four performance evaluation measures (RMSE, MAPE, R and NSEC). The results obtained in this work indicate that EEMD can effectively enhance forecasting accuracy and the proposed EEMD-ANN model can attain significant improvement over ANN approach in medium and long-term runoff time series forecasting. Copyright © 2015 Elsevier Inc. All rights reserved.
A Statistical Approach for Interval Forecasting of the Electricity Price
DEFF Research Database (Denmark)
Zhao, Jun Hua; Dong, Zhao Yang; Xu, Zhao
2008-01-01
the prediction interval is essential for estimating the uncertainty involved in the price and thus is highly useful for making generation bidding strategies and investment decisions. In this paper, a novel data mining-based approach is proposed to achieve two major objectives: 1) to accurately forecast the value......Electricity price forecasting is a difficult yet essential task for market participants in a deregulated electricity market. Rather than forecasting the value, market participants are sometimes more interested in forecasting the prediction interval of the electricity price. Forecasting...... of the electricity price series, which is widely accepted as a nonlinear time series; 2) to accurately estimate the prediction interval of the electricity price series. In the proposed approach, support vector machine (SVM) is employed to forecast the value of the price. To forecast the prediction interval, we...
Real time flood forecasting in the Upper Danube basin
Directory of Open Access Journals (Sweden)
Nester Thomas
2016-12-01
Full Text Available This paper reports on experience with developing the flood forecasting model for the Upper Danube basin and its operational use since 2006. The model system consists of hydrological and hydrodynamic components, and involves precipitation forecasts. The model parameters were estimated based on the dominant processes concept. Runoff data are assimilated in real time to update modelled soil moisture. An analysis of the model performance indicates 88% of the snow cover in the basin to be modelled correctly on more than 80% of the days. Runoff forecasting errors decrease with catchment area and increase with forecast lead time. The forecast ensemble spread is shown to be a meaningful indicator of the forecast uncertainty. During the 2013 flood, there was a tendency for the precipitation forecasts to underestimate event precipitation and for the runoff model to overestimate runoff generation which resulted in, overall, rather accurate runoff forecasts. It is suggested that the human forecaster plays an essential role in interpreting the model results and, if needed, adjusting them before issuing the forecasts to the general public.
Morrison, Kathryn T; Shaddick, Gavin; Henderson, Sarah B; Buckeridge, David L
2016-08-15
This paper outlines a latent process model for forecasting multiple health outcomes arising from a common environmental exposure. Traditionally, surveillance models in environmental health do not link health outcome measures, such as morbidity or mortality counts, to measures of exposure, such as air pollution. Moreover, different measures of health outcomes are treated as independent, while it is known that they are correlated with one another over time as they arise in part from a common underlying exposure. We propose modelling an environmental exposure as a latent process, and we describe the implementation of such a model within a hierarchical Bayesian framework and its efficient computation using integrated nested Laplace approximations. Through a simulation study, we compare distinct univariate models for each health outcome with a bivariate approach. The bivariate model outperforms the univariate models in bias and coverage of parameter estimation, in forecast accuracy and in computational efficiency. The methods are illustrated with a case study using healthcare utilization and air pollution data from British Columbia, Canada, 2003-2011, where seasonal wildfires produce high levels of air pollution, significantly impacting population health. Copyright © 2016 John Wiley & Sons, Ltd. Copyright © 2016 John Wiley & Sons, Ltd.
Short-term data forecasting based on wavelet transformation and chaos theory
Wang, Yi; Li, Cunbin; Zhang, Liang
2017-09-01
A sketch of wavelet transformation and its application was given. Concerning the characteristics of time sequence, Haar wavelet was used to do data reduction. After processing, the effect of “data nail” on forecasting was reduced. Chaos theory was also introduced, a new chaos time series forecasting flow based on wavelet transformation was proposed. The largest Lyapunov exponent was larger than zero from small data sets, it verified the data change behavior still met chaotic behavior. Based on this, chaos time series to forecast short-term change behavior could be used. At last, the example analysis of the price from a real electricity market showed that the forecasting method increased the precision of the forecasting more effectively and steadily.
A Simple Hybrid Model for Short-Term Load Forecasting
Directory of Open Access Journals (Sweden)
Suseelatha Annamareddi
2013-01-01
Full Text Available The paper proposes a simple hybrid model to forecast the electrical load data based on the wavelet transform technique and double exponential smoothing. The historical noisy load series data is decomposed into deterministic and fluctuation components using suitable wavelet coefficient thresholds and wavelet reconstruction method. The variation characteristics of the resulting series are analyzed to arrive at reasonable thresholds that yield good denoising results. The constitutive series are then forecasted using appropriate exponential adaptive smoothing models. A case study performed on California energy market data demonstrates that the proposed method can offer high forecasting precision for very short-term forecasts, considering a time horizon of two weeks.
Quality Quandaries- Time Series Model Selection and Parsimony
DEFF Research Database (Denmark)
Bisgaard, Søren; Kulahci, Murat
2009-01-01
Some of the issues involved in selecting adequate models for time series data are discussed using an example concerning the number of users of an Internet server. The process of selecting an appropriate model is subjective and requires experience and judgment. The authors believe an important...... consideration in model selection should be parameter parsimony. They favor the use of parsimonious mixed ARMA models, noting that research has shown that a model building strategy that considers only autoregressive representations will lead to non-parsimonious models and to loss of forecasting accuracy....
Tanguy, M.; Prudhomme, C.; Harrigan, S.; Smith, K. A.; Parry, S.
2017-12-01
Forecasting hydrological extremes is challenging, especially at lead times over 1 month for catchments with limited hydrological memory and variable climates. One simple way to derive monthly or seasonal hydrological forecasts is to use historical climate data to drive hydrological models using the Ensemble Streamflow Prediction (ESP) method. This gives a range of possible future streamflow given known initial hydrologic conditions alone. The degree of skill of ESP depends highly on the forecast initialisation month and catchment type. Using dynamic rainfall forecasts as driving data instead of historical data could potentially improve streamflow predictions. A lot of effort is being invested within the meteorological community to improve these forecasts. However, while recent progress shows promise (e.g. NAO in winter), the skill of these forecasts at monthly to seasonal timescales is generally still limited, and the extent to which they might lead to improved hydrological forecasts is an area of active research. Additionally, these meteorological forecasts are currently being produced at 1 month or seasonal time-steps in the UK, whereas hydrological models require forcings at daily or sub-daily time-steps. Keeping in mind these limitations of available rainfall forecasts, the objectives of this study are to find out (i) how accurate monthly dynamical rainfall forecasts need to be to outperform ESP, and (ii) how the method used to disaggregate monthly rainfall forecasts into daily rainfall time series affects results. For the first objective, synthetic rainfall time series were created by increasingly degrading observed data (proxy for a `perfect forecast') from 0 % to +/-50 % error. For the second objective, three different methods were used to disaggregate monthly rainfall data into daily time series. These were used to force a simple lumped hydrological model (GR4J) to generate streamflow predictions at a one-month lead time for over 300 catchments
Short-Term Wind Speed Forecasting for Power System Operations
Zhu, Xinxin; Genton, Marc G.
2012-01-01
some statistical short-term wind speed forecasting models, including traditional time series approaches and more advanced space-time statistical models. It also discusses the evaluation of forecast accuracy, in particular, the need for realistic loss
Knowledge fusion: An approach to time series model selection followed by pattern recognition
International Nuclear Information System (INIS)
Bleasdale, S.A.; Burr, T.L.; Scovel, J.C.; Strittmatter, R.B.
1996-03-01
This report describes work done during FY 95 that was sponsored by the Department of Energy, Office of Nonproliferation and National Security, Knowledge Fusion Project. The project team selected satellite sensor data to use as the one main example for the application of its analysis algorithms. The specific sensor-fusion problem has many generic features, which make it a worthwhile problem to attempt to solve in a general way. The generic problem is to recognize events of interest from multiple time series that define a possibly noisy background. By implementing a suite of time series modeling and forecasting methods and using well-chosen alarm criteria, we reduce the number of false alarms. We then further reduce the number of false alarms by analyzing all suspicious sections of data, as judged by the alarm criteria, with pattern recognition methods. An accompanying report (Ref 1) describes the implementation and application of this 2-step process for separating events from unusual background and applies a suite of forecasting methods followed by a suite of pattern recognition methods. This report goes into more detail about one of the forecasting methods and one of the pattern recognition methods and is applied to the same kind of satellite-sensor data that is described in Ref. 1
Long Range Dependence Prognostics for Bearing Vibration Intensity Chaotic Time Series
Directory of Open Access Journals (Sweden)
Qing Li
2016-01-01
Full Text Available According to the chaotic features and typical fractional order characteristics of the bearing vibration intensity time series, a forecasting approach based on long range dependence (LRD is proposed. In order to reveal the internal chaotic properties, vibration intensity time series are reconstructed based on chaos theory in phase-space, the delay time is computed with C-C method and the optimal embedding dimension and saturated correlation dimension are calculated via the Grassberger–Procaccia (G-P method, respectively, so that the chaotic characteristics of vibration intensity time series can be jointly determined by the largest Lyapunov exponent and phase plane trajectory of vibration intensity time series, meanwhile, the largest Lyapunov exponent is calculated by the Wolf method and phase plane trajectory is illustrated using Duffing-Holmes Oscillator (DHO. The Hurst exponent and long range dependence prediction method are proposed to verify the typical fractional order features and improve the prediction accuracy of bearing vibration intensity time series, respectively. Experience shows that the vibration intensity time series have chaotic properties and the LRD prediction method is better than the other prediction methods (largest Lyapunov, auto regressive moving average (ARMA and BP neural network (BPNN model in prediction accuracy and prediction performance, which provides a new approach for running tendency predictions for rotating machinery and provide some guidance value to the engineering practice.
Ensemble Bayesian forecasting system Part I: Theory and algorithms
Herr, Henry D.; Krzysztofowicz, Roman
2015-05-01
The ensemble Bayesian forecasting system (EBFS), whose theory was published in 2001, is developed for the purpose of quantifying the total uncertainty about a discrete-time, continuous-state, non-stationary stochastic process such as a time series of stages, discharges, or volumes at a river gauge. The EBFS is built of three components: an input ensemble forecaster (IEF), which simulates the uncertainty associated with random inputs; a deterministic hydrologic model (of any complexity), which simulates physical processes within a river basin; and a hydrologic uncertainty processor (HUP), which simulates the hydrologic uncertainty (an aggregate of all uncertainties except input). It works as a Monte Carlo simulator: an ensemble of time series of inputs (e.g., precipitation amounts) generated by the IEF is transformed deterministically through a hydrologic model into an ensemble of time series of outputs, which is next transformed stochastically by the HUP into an ensemble of time series of predictands (e.g., river stages). Previous research indicated that in order to attain an acceptable sampling error, the ensemble size must be on the order of hundreds (for probabilistic river stage forecasts and probabilistic flood forecasts) or even thousands (for probabilistic stage transition forecasts). The computing time needed to run the hydrologic model this many times renders the straightforward simulations operationally infeasible. This motivates the development of the ensemble Bayesian forecasting system with randomization (EBFSR), which takes full advantage of the analytic meta-Gaussian HUP and generates multiple ensemble members after each run of the hydrologic model; this auxiliary randomization reduces the required size of the meteorological input ensemble and makes it operationally feasible to generate a Bayesian ensemble forecast of large size. Such a forecast quantifies the total uncertainty, is well calibrated against the prior (climatic) distribution of
Time Series Data Analysis of Wireless Sensor Network Measurements of Temperature.
Bhandari, Siddhartha; Bergmann, Neil; Jurdak, Raja; Kusy, Branislav
2017-05-26
Wireless sensor networks have gained significant traction in environmental signal monitoring and analysis. The cost or lifetime of the system typically depends on the frequency at which environmental phenomena are monitored. If sampling rates are reduced, energy is saved. Using empirical datasets collected from environmental monitoring sensor networks, this work performs time series analyses of measured temperature time series. Unlike previous works which have concentrated on suppressing the transmission of some data samples by time-series analysis but still maintaining high sampling rates, this work investigates reducing the sampling rate (and sensor wake up rate) and looks at the effects on accuracy. Results show that the sampling period of the sensor can be increased up to one hour while still allowing intermediate and future states to be estimated with interpolation RMSE less than 0.2 °C and forecasting RMSE less than 1 °C.
Selecting the Best Forecasting-Implied Volatility Model Using Genetic Programming
Directory of Open Access Journals (Sweden)
Wafa Abdelmalek
2009-01-01
Full Text Available The volatility is a crucial variable in option pricing and hedging strategies. The aim of this paper is to provide some initial evidence of the empirical relevance of genetic programming to volatility's forecasting. By using real data from S&P500 index options, the genetic programming's ability to forecast Black and Scholes-implied volatility is compared between time series samples and moneyness-time to maturity classes. Total and out-of-sample mean squared errors are used as forecasting's performance measures. Comparisons reveal that the time series model seems to be more accurate in forecasting-implied volatility than moneyness time to maturity models. Overall, results are strongly encouraging and suggest that the genetic programming approach works well in solving financial problems.
Barbetta, Silvia; Coccia, Gabriele; Moramarco, Tommaso; Brocca, Luca; Todini, Ezio
2017-08-01
This work extends the multi-temporal approach of the Model Conditional Processor (MCP-MT) to the multi-model case and to the four Truncated Normal Distributions (TNDs) approach, demonstrating the improvement on the single-temporal one. The study is framed in the context of probabilistic Bayesian decision-making that is appropriate to take rational decisions on uncertain future outcomes. As opposed to the direct use of deterministic forecasts, the probabilistic forecast identifies a predictive probability density function that represents a fundamental knowledge on future occurrences. The added value of MCP-MT is the identification of the probability that a critical situation will happen within the forecast lead-time and when, more likely, it will occur. MCP-MT is thoroughly tested for both single-model and multi-model configurations at a gauged site on the Tiber River, central Italy. The stages forecasted by two operative deterministic models, STAFOM-RCM and MISDc, are considered for the study. The dataset used for the analysis consists of hourly data from 34 flood events selected on a time series of six years. MCP-MT improves over the original models' forecasts: the peak overestimation and the rising limb delayed forecast, characterizing MISDc and STAFOM-RCM respectively, are significantly mitigated, with a reduced mean error on peak stage from 45 to 5 cm and an increased coefficient of persistence from 0.53 up to 0.75. The results show that MCP-MT outperforms the single-temporal approach and is potentially useful for supporting decision-making because the exceedance probability of hydrometric thresholds within a forecast horizon and the most probable flooding time can be estimated.
Real-time Social Internet Data to Guide Forecasting Models
Energy Technology Data Exchange (ETDEWEB)
Del Valle, Sara Y. [Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
2016-09-20
Our goal is to improve decision support by monitoring and forecasting events using social media, mathematical models, and quantifying model uncertainty. Our approach is real-time, data-driven forecasts with quantified uncertainty: Not just for weather anymore. Information flow from human observations of events through an Internet system and classification algorithms is used to produce quantitatively uncertain forecast. In summary, we want to develop new tools to extract useful information from Internet data streams, develop new approaches to assimilate real-time information into predictive models, validate approaches by forecasting events, and our ultimate goal is to develop an event forecasting system using mathematical approaches and heterogeneous data streams.
Real-time flood inundation forecasting and mapping for key railway infrastructure: a UK case study
Directory of Open Access Journals (Sweden)
Murphy Alexandra T.
2016-01-01
Full Text Available Flooding events that impede railway infrastructure can cause severe travel delays for the general public and large fines in delayed minutes for the rail industry. Early warnings of flood inundation can give more time to implement mitigation measures which help reduce cancellations, delays and fines. Initial work is reported on the development of a real-time flood inundation forecasting and mapping system for the Cowley Bridge track area near Exeter, UK. This location is on one of the main access routes to South West England and has suffered major floods in the past resulting in significant transport impacts. Flood forecasting systems in the UK mainly forecast river level/flow rather than extent and depth of flood inundation. Here, the development of a chain of coupled models is discussed that link rainfall to river flow, river level and flood extent for the rail track area relating to Cowley Bridge. Historical events are identified to test model performance in predicting inundation of railway infrastructure. The modelling system will operate alongside a series of in-situ sensors chosen to enhance the flood mapping forecasting system. Sensor data will support offline model calibration/verification and real-time data assimilation as well as monitoring flood conditions to inform track closure decisions.
Predicting the Market Potential Using Time Series Analysis
Directory of Open Access Journals (Sweden)
Halmet Bradosti
2015-12-01
Full Text Available The aim of this analysis is to forecast a mini-market sales volume for the period of twelve months starting August 2015 to August 2016. The study is based on the monthly sales in Iraqi Dinar for a private local mini-market for the month of April 2014 to July 2015. As revealed on the graph and of course if the stagnant economic condition continues, the trend of future sales is down-warding. Based on time series analysis, the business may continue to operate and generate small revenues until August 2016. However, due to low sales volume, low profit margin and operating expenses, the revenues may not be adequate enough to produce positive net income and the business may not be able to operate afterward. The principal question rose from this is the forecasting sales in the region will be difficult where the business cycle so dynamic and revolutionary due to systematic risks and unforeseeable future.
Applications of soft computing in time series forecasting simulation and modeling techniques
Singh, Pritpal
2016-01-01
This book reports on an in-depth study of fuzzy time series (FTS) modeling. It reviews and summarizes previous research work in FTS modeling and also provides a brief introduction to other soft-computing techniques, such as artificial neural networks (ANNs), rough sets (RS) and evolutionary computing (EC), focusing on how these techniques can be integrated into different phases of the FTS modeling approach. In particular, the book describes novel methods resulting from the hybridization of FTS modeling approaches with neural networks and particle swarm optimization. It also demonstrates how a new ANN-based model can be successfully applied in the context of predicting Indian summer monsoon rainfall. Thanks to its easy-to-read style and the clear explanations of the models, the book can be used as a concise yet comprehensive reference guide to fuzzy time series modeling, and will be valuable not only for graduate students, but also for researchers and professionals working for academic, business and governmen...
Day-ahead electricity price forecasting using wavelet transform combined with ARIMA and GARCH models
International Nuclear Information System (INIS)
Tan, Zhongfu; Zhang, Jinliang; Xu, Jun; Wang, Jianhui
2010-01-01
This paper proposes a novel price forecasting method based on wavelet transform combined with ARIMA and GARCH models. By wavelet transform, the historical price series is decomposed and reconstructed into one approximation series and some detail series. Then each subseries can be separately predicted by a suitable time series model. The final forecast is obtained by composing the forecasted results of each subseries. This proposed method is examined on Spanish and PJM electricity markets and compared with some other forecasting methods. (author)
Verification of short lead time forecast models: applied to Kp and Dst forecasting
Wintoft, Peter; Wik, Magnus
2016-04-01
In the ongoing EU/H2020 project PROGRESS models that predicts Kp, Dst, and AE from L1 solar wind data will be used as inputs to radiation belt models. The possible lead times from L1 measurements are shorter (10s of minutes to hours) than the typical duration of the physical phenomena that should be forecast. Under these circumstances several metrics fail to single out trivial cases, such as persistence. In this work we explore metrics and approaches for short lead time forecasts. We apply these to current Kp and Dst forecast models. This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 637302.
Computational intelligence in time series forecasting theory and engineering applications
Palit, Ajoy K
2005-01-01
Foresight in an engineering enterprise can make the difference between success and failure, and can be vital to the effective control of industrial systems. Applying time series analysis in the on-line milieu of most industrial plants has been problematic owing to the time and computational effort required. The advent of soft computing tools offers a solution. The authors harness the power of intelligent technologies individually and in combination. Examples of the particular systems and processes susceptible to each technique are investigated, cultivating a comprehensive exposition of the improvements on offer in quality, model building and predictive control and the selection of appropriate tools from the plethora available. Application-oriented engineers in process control, manufacturing, production industry and research centres will find much to interest them in this book. It is suitable for industrial training purposes, as well as serving as valuable reference material for experimental researchers.
Time series analysis of reference crop evapotranspiration for Bokaro District, Jharkhand, India
Directory of Open Access Journals (Sweden)
Gautam Ratnesh
2016-09-01
Full Text Available Evapotranspiration is the one of the major role playing element in water cycle. More accurate measurement and forecasting of Evapotranspiration would enable more efficient water resources management. This study, is therefore, particularly focused on evapotranspiration modelling and forecasting, since forecasting would provide better information for optimal water resources management. There are numerous techniques of evapotranspiration forecasting that include autoregressive (AR and moving average (MA, autoregressive moving average (ARMA, autoregressive integrated moving average (ARIMA, Thomas Feiring, etc. Out of these models ARIMA model has been found to be more suitable for analysis and forecasting of hydrological events. Therefore, in this study ARIMA models have been used for forecasting of mean monthly reference crop evapotranspiration by stochastic analysis. The data series of 102 years i.e. 1224 months of Bokaro District were used for analysis and forecasting. Different order of ARIMA model was selected on the basis of autocorrelation function (ACF and partial autocorrelation (PACF of data series. Maximum likelihood method was used for determining the parameters of the models. To see the statistical parameter of model, best fitted model is ARIMA (0, 1, 4 (0, 1, 112.
Price forecasting of day-ahead electricity markets using a hybrid forecast method
International Nuclear Information System (INIS)
Shafie-khah, M.; Moghaddam, M. Parsa; Sheikh-El-Eslami, M.K.
2011-01-01
Research highlights: → A hybrid method is proposed to forecast the day-ahead prices in electricity market. → The method combines Wavelet-ARIMA and RBFN network models. → PSO method is applied to obtain optimum RBFN structure for avoiding over fitting. → One of the merits of the proposed method is lower need to the input data. → The proposed method has more accurate behavior in compare with previous methods. -- Abstract: Energy price forecasting in a competitive electricity market is crucial for the market participants in planning their operations and managing their risk, and it is also the key information in the economic optimization of the electric power industry. However, price series usually have a complex behavior due to their nonlinearity, nonstationarity, and time variancy. In this paper, a novel hybrid method to forecast day-ahead electricity price is proposed. This hybrid method is based on wavelet transform, Auto-Regressive Integrated Moving Average (ARIMA) models and Radial Basis Function Neural Networks (RBFN). The wavelet transform provides a set of better-behaved constitutive series than price series for prediction. ARIMA model is used to generate a linear forecast, and then RBFN is developed as a tool for nonlinear pattern recognition to correct the estimation error in wavelet-ARIMA forecast. Particle Swarm Optimization (PSO) is used to optimize the network structure which makes the RBFN be adapted to the specified training set, reducing computation complexity and avoiding overfitting. The proposed method is examined on the electricity market of mainland Spain and the results are compared with some of the most recent price forecast methods. The results show that the proposed hybrid method could provide a considerable improvement for the forecasting accuracy.
Price forecasting of day-ahead electricity markets using a hybrid forecast method
Energy Technology Data Exchange (ETDEWEB)
Shafie-khah, M., E-mail: miadreza@gmail.co [Tarbiat Modares University, Tehran (Iran, Islamic Republic of); Moghaddam, M. Parsa, E-mail: parsa@modares.ac.i [Tarbiat Modares University, Tehran (Iran, Islamic Republic of); Sheikh-El-Eslami, M.K., E-mail: aleslam@modares.ac.i [Tarbiat Modares University, Tehran (Iran, Islamic Republic of)
2011-05-15
Research highlights: {yields} A hybrid method is proposed to forecast the day-ahead prices in electricity market. {yields} The method combines Wavelet-ARIMA and RBFN network models. {yields} PSO method is applied to obtain optimum RBFN structure for avoiding over fitting. {yields} One of the merits of the proposed method is lower need to the input data. {yields} The proposed method has more accurate behavior in compare with previous methods. -- Abstract: Energy price forecasting in a competitive electricity market is crucial for the market participants in planning their operations and managing their risk, and it is also the key information in the economic optimization of the electric power industry. However, price series usually have a complex behavior due to their nonlinearity, nonstationarity, and time variancy. In this paper, a novel hybrid method to forecast day-ahead electricity price is proposed. This hybrid method is based on wavelet transform, Auto-Regressive Integrated Moving Average (ARIMA) models and Radial Basis Function Neural Networks (RBFN). The wavelet transform provides a set of better-behaved constitutive series than price series for prediction. ARIMA model is used to generate a linear forecast, and then RBFN is developed as a tool for nonlinear pattern recognition to correct the estimation error in wavelet-ARIMA forecast. Particle Swarm Optimization (PSO) is used to optimize the network structure which makes the RBFN be adapted to the specified training set, reducing computation complexity and avoiding overfitting. The proposed method is examined on the electricity market of mainland Spain and the results are compared with some of the most recent price forecast methods. The results show that the proposed hybrid method could provide a considerable improvement for the forecasting accuracy.
Energy Technology Data Exchange (ETDEWEB)
Neto, Joao C. do L, E-mail: jcaldas@ufam.edu.br [Group of Optimization and Fuzzy Systems, Federal University of Amazonas, General Rodrigo Octavio Jordao Ramos Avenue, 3000, Academic Campus, 69077-000 Manaus, Amazonas (Brazil); Costa Junior, Carlos T. da [Postgraduate Program in Electrical Engineering, Institute of Technology, Federal University of Para, Augusto Correa Street, 1, Guama, 66075-900 Belem, Para (Brazil); Bitar, Sandro D.B. [Group of Optimization and Fuzzy Systems, Federal University of Amazonas, General Rodrigo Octavio Jordao Ramos Avenue, 3000, Academic Campus, 69077-000 Manaus, Amazonas (Brazil); Junior, Walter B. [Postgraduate Program in Electrical Engineering, Institute of Technology, Federal University of Para, Augusto Correa Street, 1, Guama, 66075-900 Belem, Para (Brazil)
2011-09-15
Understanding the uncertainty inherent in the analysis of diesel fuel consumption and its impact on the generation of electricity is an important topic for planning the expansion of isolated thermoelectric systems in the state of Amazonas. In light of this, a decision support system has been developed to forecast the cost of electricity production using non-stationary data by integrating the methodology of time series models with fuzzy systems and optimization tools. The method presented herein combines the potential of the Autoregressive Integrated Moving Average (ARIMA) and the Seasonal ARIMA (SARIMA) models, such as the forecasting tool, with the advantages of fuzzy set theory to compensate for the uncertainties and errors encountered in the observed data, which would degrade the validity of forecasted values. The results show that incorporation of the {alpha}-cut concept facilitated the evaluation of risks while allowing simultaneous consideration of intervals for the unitary cost of energy production. This provides the analyst with the ability to make decisions using various predicted intervals with different membership values instead of the common practice of simply using the specific costs. - Highlights: > A decision support system has been developed using SARIMA with fuzzy systems and optimizations tools. > It assists the decision-making process for planning the expansion in isolated thermoelectric systems. > The {alpha}-cut concept facilitated the evaluation of risks for the cost of electricity production. > Provides decisions using various forecasted interval for this cost with different membership values.
International Nuclear Information System (INIS)
Neto, Joao C. do L; Costa Junior, Carlos T. da; Bitar, Sandro D.B.; Junior, Walter B.
2011-01-01
Understanding the uncertainty inherent in the analysis of diesel fuel consumption and its impact on the generation of electricity is an important topic for planning the expansion of isolated thermoelectric systems in the state of Amazonas. In light of this, a decision support system has been developed to forecast the cost of electricity production using non-stationary data by integrating the methodology of time series models with fuzzy systems and optimization tools. The method presented herein combines the potential of the Autoregressive Integrated Moving Average (ARIMA) and the Seasonal ARIMA (SARIMA) models, such as the forecasting tool, with the advantages of fuzzy set theory to compensate for the uncertainties and errors encountered in the observed data, which would degrade the validity of forecasted values. The results show that incorporation of the α-cut concept facilitated the evaluation of risks while allowing simultaneous consideration of intervals for the unitary cost of energy production. This provides the analyst with the ability to make decisions using various predicted intervals with different membership values instead of the common practice of simply using the specific costs. - Highlights: → A decision support system has been developed using SARIMA with fuzzy systems and optimizations tools. → It assists the decision-making process for planning the expansion in isolated thermoelectric systems. → The α-cut concept facilitated the evaluation of risks for the cost of electricity production. → Provides decisions using various forecasted interval for this cost with different membership values.
Wind and load forecast error model for multiple geographically distributed forecasts
Energy Technology Data Exchange (ETDEWEB)
Makarov, Yuri V.; Reyes-Spindola, Jorge F.; Samaan, Nader; Diao, Ruisheng; Hafen, Ryan P. [Pacific Northwest National Laboratory, Richland, WA (United States)
2010-07-01
The impact of wind and load forecast errors on power grid operations is frequently evaluated by conducting multi-variant studies, where these errors are simulated repeatedly as random processes based on their known statistical characteristics. To simulate these errors correctly, we need to reflect their distributions (which do not necessarily follow a known distribution law), standard deviations. auto- and cross-correlations. For instance, load and wind forecast errors can be closely correlated in different zones of the system. This paper introduces a new methodology for generating multiple cross-correlated random processes to produce forecast error time-domain curves based on a transition probability matrix computed from an empirical error distribution function. The matrix will be used to generate new error time series with statistical features similar to observed errors. We present the derivation of the method and some experimental results obtained by generating new error forecasts together with their statistics. (orig.)
Financial Time Series Prediction Using Elman Recurrent Random Neural Networks
Directory of Open Access Journals (Sweden)
Jie Wang
2016-01-01
(ERNN, the empirical results show that the proposed neural network displays the best performance among these neural networks in financial time series forecasting. Further, the empirical research is performed in testing the predictive effects of SSE, TWSE, KOSPI, and Nikkei225 with the established model, and the corresponding statistical comparisons of the above market indices are also exhibited. The experimental results show that this approach gives good performance in predicting the values from the stock market indices.
Forecasting interest rates with shifting endpoints
DEFF Research Database (Denmark)
Van Dijk, Dick; Koopman, Siem Jan; Wel, Michel van der
2014-01-01
We consider forecasting the term structure of interest rates with the assumption that factors driving the yield curve are stationary around a slowly time-varying mean or ‘shifting endpoint’. The shifting endpoints are captured using either (i) time series methods (exponential smoothing) or (ii......) long-range survey forecasts of either interest rates or inflation and output growth, or (iii) exponentially smoothed realizations of these macro variables. Allowing for shifting endpoints in yield curve factors provides substantial and significant gains in out-of-sample predictive accuracy, relative...... to stationary and random walk benchmarks. Forecast improvements are largest for long-maturity interest rates and for long-horizon forecasts....
Work-related accidents among the Iranian population: a time series analysis, 2000-2011.
Karimlou, Masoud; Salehi, Masoud; Imani, Mehdi; Hosseini, Agha-Fatemeh; Dehnad, Afsaneh; Vahabi, Nasim; Bakhtiyari, Mahmood
2015-01-01
Work-related accidents result in human suffering and economic losses and are considered as a major health problem worldwide, especially in the economically developing world. To introduce seasonal autoregressive moving average (ARIMA) models for time series analysis of work-related accident data for workers insured by the Iranian Social Security Organization (ISSO) between 2000 and 2011. In this retrospective study, all insured people experiencing at least one work-related accident during a 10-year period were included in the analyses. We used Box-Jenkins modeling to develop a time series model of the total number of accidents. There was an average of 1476 accidents per month (1476·05±458·77, mean±SD). The final ARIMA (p,d,q) (P,D,Q)s model for fitting to data was: ARIMA(1,1,1)×(0,1,1)12 consisting of the first ordering of the autoregressive, moving average and seasonal moving average parameters with 20·942 mean absolute percentage error (MAPE). The final model showed that time series analysis of ARIMA models was useful for forecasting the number of work-related accidents in Iran. In addition, the forecasted number of work-related accidents for 2011 explained the stability of occurrence of these accidents in recent years, indicating a need for preventive occupational health and safety policies such as safety inspection.
A novel hybrid ensemble learning paradigm for tourism forecasting
Shabri, Ani
2015-02-01
In this paper, a hybrid forecasting model based on Empirical Mode Decomposition (EMD) and Group Method of Data Handling (GMDH) is proposed to forecast tourism demand. This methodology first decomposes the original visitor arrival series into several Intrinsic Model Function (IMFs) components and one residual component by EMD technique. Then, IMFs components and the residual components is forecasted respectively using GMDH model whose input variables are selected by using Partial Autocorrelation Function (PACF). The final forecasted result for tourism series is produced by aggregating all the forecasted results. For evaluating the performance of the proposed EMD-GMDH methodologies, the monthly data of tourist arrivals from Singapore to Malaysia are used as an illustrative example. Empirical results show that the proposed EMD-GMDH model outperforms the EMD-ARIMA as well as the GMDH and ARIMA (Autoregressive Integrated Moving Average) models without time series decomposition.
Assessing energy forecasting inaccuracy by simultaneously considering temporal and absolute errors
International Nuclear Information System (INIS)
Frías-Paredes, Laura; Mallor, Fermín; Gastón-Romeo, Martín; León, Teresa
2017-01-01
Highlights: • A new method to match time series is defined to assess energy forecasting accuracy. • This method relies in a new family of step patterns that optimizes the MAE. • A new definition of the Temporal Distortion Index between two series is provided. • A parametric extension controls both the temporal distortion index and the MAE. • Pareto optimal transformations of the forecast series are obtained for both indexes. - Abstract: Recent years have seen a growing trend in wind and solar energy generation globally and it is expected that an important percentage of total energy production comes from these energy sources. However, they present inherent variability that implies fluctuations in energy generation that are difficult to forecast. Thus, forecasting errors have a considerable role in the impacts and costs of renewable energy integration, management, and commercialization. This study presents an important advance in the task of analyzing prediction models, in particular, in the timing component of prediction error, which improves previous pioneering results. A new method to match time series is defined in order to assess energy forecasting accuracy. This method relies on a new family of step patterns, an essential component of the algorithm to evaluate the temporal distortion index (TDI). This family minimizes the mean absolute error (MAE) of the transformation with respect to the reference series (the real energy series) and also allows detailed control of the temporal distortion entailed in the prediction series. The simultaneous consideration of temporal and absolute errors allows the use of Pareto frontiers as characteristic error curves. Real examples of wind energy forecasts are used to illustrate the results.
Combining Kohonen maps with Arima time series models to forecast traffic flow
van der Voort, Mascha C.; Dougherty, Mark; Dougherty, M.S.; Watson, Susan
1996-01-01
A hybrid method of short-term traffic forecasting is introduced; the KARIMA method. The technique uses a Kohonen self-organizing map as an initial classifier; each class has an individually tuned ARIMA model associated with it. Using a Kohonen map which is hexagonal in layout eases the problem of
Price Density Forecasts in the U.S. Hog Market: Composite Procedures
Trujillo Barrera, A.A.; Garcia, P.; Mallory, M.
2013-01-01
Abstract We develop and evaluate quarterly out-of-sample individual and composite density forecasts for U.S. hog prices using data from 1975.I to 2010.IV. Individual forecasts are generated from time series models and the implied distribution of USDA outlook forecasts. Composite density forecasts
Kadri, Farid; Harrou, Fouzi; Sun, Ying
2018-01-01
Efficient management of patient demands in emergency departments (EDs) has recently received increasing attention by most healthcare administrations. Forecasting ED demands greatly helps ED's managers to make suitable decisions by optimally
Power Forecasting of Combined Heating and Cooling Systems Based on Chaotic Time Series
Directory of Open Access Journals (Sweden)
Liu Hai
2015-01-01
Full Text Available Theoretic analysis shows that the output power of the distributed generation system is nonlinear and chaotic. And it is coupled with the microenvironment meteorological data. Chaos is an inherent property of nonlinear dynamic system. A predicator of the output power of the distributed generation system is to establish a nonlinear model of the dynamic system based on real time series in the reconstructed phase space. Firstly, chaos should be detected and quantified for the intensive studies of nonlinear systems. If the largest Lyapunov exponent is positive, the dynamical system must be chaotic. Then, the embedding dimension and the delay time are chosen based on the improved C-C method. The attractor of chaotic power time series can be reconstructed based on the embedding dimension and delay time in the phase space. By now, the neural network can be trained based on the training samples, which are observed from the distributed generation system. The neural network model will approximate the curve of output power adequately. Experimental results show that the maximum power point of the distributed generation system will be predicted based on the meteorological data. The system can be controlled effectively based on the prediction.
Winter Holts Oscillatory Method: A New Method of Resampling in Time Series.
Directory of Open Access Journals (Sweden)
Muhammad Imtiaz Subhani
2016-12-01
Full Text Available The core proposition behind this research is to create innovative methods of bootstrapping that can be applied in time series data. In order to find new methods of bootstrapping, various methods were reviewed; The data of automotive Sales, Market Shares and Net Exports of the top 10 countries, which includes China, Europe, United States of America (USA, Japan, Germany, South Korea, India, Mexico, Brazil, Spain and, Canada from 2002 to 2014 were collected through various sources which includes UN Comtrade, Index Mundi and World Bank. The findings of this paper confirmed that Bootstrapping for resampling through winter forecasting by Oscillation and Average methods give more robust results than the winter forecasting by any general methods.
Interpretation of a compositional time series
Tolosana-Delgado, R.; van den Boogaart, K. G.
2012-04-01
Common methods for multivariate time series analysis use linear operations, from the definition of a time-lagged covariance/correlation to the prediction of new outcomes. However, when the time series response is a composition (a vector of positive components showing the relative importance of a set of parts in a total, like percentages and proportions), then linear operations are afflicted of several problems. For instance, it has been long recognised that (auto/cross-)correlations between raw percentages are spurious, more dependent on which other components are being considered than on any natural link between the components of interest. Also, a long-term forecast of a composition in models with a linear trend will ultimately predict negative components. In general terms, compositional data should not be treated in a raw scale, but after a log-ratio transformation (Aitchison, 1986: The statistical analysis of compositional data. Chapman and Hill). This is so because the information conveyed by a compositional data is relative, as stated in their definition. The principle of working in coordinates allows to apply any sort of multivariate analysis to a log-ratio transformed composition, as long as this transformation is invertible. This principle is of full application to time series analysis. We will discuss how results (both auto/cross-correlation functions and predictions) can be back-transformed, viewed and interpreted in a meaningful way. One view is to use the exhaustive set of all possible pairwise log-ratios, which allows to express the results into D(D - 1)/2 separate, interpretable sets of one-dimensional models showing the behaviour of each possible pairwise log-ratios. Another view is the interpretation of estimated coefficients or correlations back-transformed in terms of compositions. These two views are compatible and complementary. These issues are illustrated with time series of seasonal precipitation patterns at different rain gauges of the USA
Wavelet regression model in forecasting crude oil price
Hamid, Mohd Helmie; Shabri, Ani
2017-05-01
This study presents the performance of wavelet multiple linear regression (WMLR) technique in daily crude oil forecasting. WMLR model was developed by integrating the discrete wavelet transform (DWT) and multiple linear regression (MLR) model. The original time series was decomposed to sub-time series with different scales by wavelet theory. Correlation analysis was conducted to assist in the selection of optimal decomposed components as inputs for the WMLR model. The daily WTI crude oil price series has been used in this study to test the prediction capability of the proposed model. The forecasting performance of WMLR model were also compared with regular multiple linear regression (MLR), Autoregressive Moving Average (ARIMA) and Generalized Autoregressive Conditional Heteroscedasticity (GARCH) using root mean square errors (RMSE) and mean absolute errors (MAE). Based on the experimental results, it appears that the WMLR model performs better than the other forecasting technique tested in this study.
Chaos Time Series Prediction Based on Membrane Optimization Algorithms
Directory of Open Access Journals (Sweden)
Meng Li
2015-01-01
Full Text Available This paper puts forward a prediction model based on membrane computing optimization algorithm for chaos time series; the model optimizes simultaneously the parameters of phase space reconstruction (τ,m and least squares support vector machine (LS-SVM (γ,σ by using membrane computing optimization algorithm. It is an important basis for spectrum management to predict accurately the change trend of parameters in the electromagnetic environment, which can help decision makers to adopt an optimal action. Then, the model presented in this paper is used to forecast band occupancy rate of frequency modulation (FM broadcasting band and interphone band. To show the applicability and superiority of the proposed model, this paper will compare the forecast model presented in it with conventional similar models. The experimental results show that whether single-step prediction or multistep prediction, the proposed model performs best based on three error measures, namely, normalized mean square error (NMSE, root mean square error (RMSE, and mean absolute percentage error (MAPE.
Linear stochastic models for forecasting daily maxima and hourly concentrations of air pollutants
Energy Technology Data Exchange (ETDEWEB)
McCollister, G M; Wilson, K R
1975-04-01
Two related time series models were developed to forecast concentrations of various air pollutants and tested on carbon monoxide and oxidant data for the Los Angeles basin. One model forecasts daily maximum concentrations of a particular pollutant using only past daily maximum values of that pollutant as input. The other model forecasts 1 hr average concentrations using only the past hourly average values. Both are significantly more accurate than persistence, i.e., forecasting for tomorrow what occurred today (or yesterday). Model forecasts for 1972 of the daily instantaneous maxima for total oxidant made using only past pollutant concentration data are more accurate than those made by the Los Angeles APCD using meteorological input as well as pollutant concentrations. Although none of these models forecast as accurately as might be desired for a health warning system, the relative success of simple time series models, even though based solely on pollutant concentration, suggests that models incorporating meteorological data and using either multi-dimensional times series or pattern recognition techniques should be tested.
Forecast of useful energy for the TIMES-Norway model
International Nuclear Information System (INIS)
Rosenberg, Eva
2012-01-01
A regional forecast of useful energy demand in seven Norwegian regions is calculated based on an earlier work with a national forecast. This forecast will be input to the energy system model TIMES-Norway and analyses will result in forecasts of energy use of different energy carriers with varying external conditions (not included in this report). The forecast presented here describes the methodology used and the resulting forecast of useful energy. lt is based on information of the long-term development of the economy by the Ministry of Finance, projections of population growths from Statistics Norway and several other studies. The definition of a forecast of useful energy demand is not absolute, but depends on the purpose. One has to be careful not to include parts that are a part of the energy system model, such as energy efficiency measures. In the forecast presented here the influence of new building regulations and the prohibition of production of incandescent light bulbs in EU etc. are included. Other energy efficiency measures such as energy management, heat pumps, tightening of leaks etc. are modelled as technologies to invest in and are included in the TIMES-Norway model. The elasticity between different energy carriers are handled by the TIMES-Norway model and some elasticity is also included as the possibility to invest in energy efficiency measures. The forecast results in an increase of the total useful energy from 2006 to 2050 by 18 o/o. The growth is expected to be highest in the regions South and East. The industry remains at a constant level in the base case and increased or reduced energy demand is analysed as different scenarios with the TIMES-Norway model. The most important driver is the population growth. Together with the assumptions made it results in increased useful energy demand in the household and service sectors of 25 o/o and 57 % respectively.(au)
Forecast of useful energy for the TIMES-Norway model
Energy Technology Data Exchange (ETDEWEB)
Rosenberg, Eva
2012-07-25
A regional forecast of useful energy demand in seven Norwegian regions is calculated based on an earlier work with a national forecast. This forecast will be input to the energy system model TIMES-Norway and analyses will result in forecasts of energy use of different energy carriers with varying external conditions (not included in this report). The forecast presented here describes the methodology used and the resulting forecast of useful energy. lt is based on information of the long-term development of the economy by the Ministry of Finance, projections of population growths from Statistics Norway and several other studies. The definition of a forecast of useful energy demand is not absolute, but depends on the purpose. One has to be careful not to include parts that are a part of the energy system model, such as energy efficiency measures. In the forecast presented here the influence of new building regulations and the prohibition of production of incandescent light bulbs in EU etc. are included. Other energy efficiency measures such as energy management, heat pumps, tightening of leaks etc. are modelled as technologies to invest in and are included in the TIMES-Norway model. The elasticity between different energy carriers are handled by the TIMES-Norway model and some elasticity is also included as the possibility to invest in energy efficiency measures. The forecast results in an increase of the total useful energy from 2006 to 2050 by 18 o/o. The growth is expected to be highest in the regions South and East. The industry remains at a constant level in the base case and increased or reduced energy demand is analysed as different scenarios with the TIMES-Norway model. The most important driver is the population growth. Together with the assumptions made it results in increased useful energy demand in the household and service sectors of 25 o/o and 57 % respectively.(au)
Assimilation of LAI time-series in crop production models
Kooistra, Lammert; Rijk, Bert; Nannes, Louis
2014-05-01
Agriculture is worldwide a large consumer of freshwater, nutrients and land. Spatial explicit agricultural management activities (e.g., fertilization, irrigation) could significantly improve efficiency in resource use. In previous studies and operational applications, remote sensing has shown to be a powerful method for spatio-temporal monitoring of actual crop status. As a next step, yield forecasting by assimilating remote sensing based plant variables in crop production models would improve agricultural decision support both at the farm and field level. In this study we investigated the potential of remote sensing based Leaf Area Index (LAI) time-series assimilated in the crop production model LINTUL to improve yield forecasting at field level. The effect of assimilation method and amount of assimilated observations was evaluated. The LINTUL-3 crop production model was calibrated and validated for a potato crop on two experimental fields in the south of the Netherlands. A range of data sources (e.g., in-situ soil moisture and weather sensors, destructive crop measurements) was used for calibration of the model for the experimental field in 2010. LAI from cropscan field radiometer measurements and actual LAI measured with the LAI-2000 instrument were used as input for the LAI time-series. The LAI time-series were assimilated in the LINTUL model and validated for a second experimental field on which potatoes were grown in 2011. Yield in 2011 was simulated with an R2 of 0.82 when compared with field measured yield. Furthermore, we analysed the potential of assimilation of LAI into the LINTUL-3 model through the 'updating' assimilation technique. The deviation between measured and simulated yield decreased from 9371 kg/ha to 8729 kg/ha when assimilating weekly LAI measurements in the LINTUL model over the season of 2011. LINTUL-3 furthermore shows the main growth reducing factors, which are useful for farm decision support. The combination of crop models and sensor
Essays in real-time forecasting
Liebermann, Joelle
2012-01-01
This thesis contains three essays in the field of real-time econometrics, and more particularlyforecasting.The issue of using data as available in real-time to forecasters, policymakers or financialmarkets is an important one which has only recently been taken on board in the empiricalliterature. Data available and used in real-time are preliminary and differ from ex-postrevised data, and given that data revisions may be quite substantial, the use of latestavailable instead of real-time can s...
Time series analysis of temporal networks
Sikdar, Sandipan; Ganguly, Niloy; Mukherjee, Animesh
2016-01-01
A common but an important feature of all real-world networks is that they are temporal in nature, i.e., the network structure changes over time. Due to this dynamic nature, it becomes difficult to propose suitable growth models that can explain the various important characteristic properties of these networks. In fact, in many application oriented studies only knowing these properties is sufficient. For instance, if one wishes to launch a targeted attack on a network, this can be done even without the knowledge of the full network structure; rather an estimate of some of the properties is sufficient enough to launch the attack. We, in this paper show that even if the network structure at a future time point is not available one can still manage to estimate its properties. We propose a novel method to map a temporal network to a set of time series instances, analyze them and using a standard forecast model of time series, try to predict the properties of a temporal network at a later time instance. To our aim, we consider eight properties such as number of active nodes, average degree, clustering coefficient etc. and apply our prediction framework on them. We mainly focus on the temporal network of human face-to-face contacts and observe that it represents a stochastic process with memory that can be modeled as Auto-Regressive-Integrated-Moving-Average (ARIMA). We use cross validation techniques to find the percentage accuracy of our predictions. An important observation is that the frequency domain properties of the time series obtained from spectrogram analysis could be used to refine the prediction framework by identifying beforehand the cases where the error in prediction is likely to be high. This leads to an improvement of 7.96% (for error level ≤20%) in prediction accuracy on an average across all datasets. As an application we show how such prediction scheme can be used to launch targeted attacks on temporal networks. Contribution to the Topical Issue
International Nuclear Information System (INIS)
Liu, Xiaoyong; Fu, Hui
2016-01-01
Interbank Offered rate is the only direct market rate in China’s currency market. Volatility forecasting of China Interbank Offered Rate (IBOR) has a very important theoretical and practical significance for financial asset pricing and financial risk measure or management. However, IBOR is a dynamics and non-steady time series whose developmental changes have stronger random fluctuation, so it is difficult to forecast the volatility of IBOR. This paper offers a hybrid algorithm using grey model and extreme learning machine (ELM) to forecast volatility of IBOR. The proposed algorithm is composed of three phases. In the first, grey model is used to deal with the original IBOR time series by accumulated generating operation (AGO) and weaken the stochastic volatility in original series. And then, a forecasting model is founded by using ELM to analyze the new IBOR series. Lastly, the predictive value of the original IBOR series can be obtained by inverse accumulated generating operation (IAGO). The new model is applied to forecasting Interbank Offered Rate of China. Compared with the forecasting results of BP and classical ELM, the new model is more efficient to forecasting short- and middle-term volatility of IBOR.
Work-related accidents among the Iranian population: a time series analysis, 2000–2011
Karimlou, Masoud; Imani, Mehdi; Hosseini, Agha-Fatemeh; Dehnad, Afsaneh; Vahabi, Nasim; Bakhtiyari, Mahmood
2015-01-01
Background Work-related accidents result in human suffering and economic losses and are considered as a major health problem worldwide, especially in the economically developing world. Objectives To introduce seasonal autoregressive moving average (ARIMA) models for time series analysis of work-related accident data for workers insured by the Iranian Social Security Organization (ISSO) between 2000 and 2011. Methods In this retrospective study, all insured people experiencing at least one work-related accident during a 10-year period were included in the analyses. We used Box–Jenkins modeling to develop a time series model of the total number of accidents. Results There was an average of 1476 accidents per month (1476·05±458·77, mean±SD). The final ARIMA (p,d,q) (P,D,Q)s model for fitting to data was: ARIMA(1,1,1)×(0,1,1)12 consisting of the first ordering of the autoregressive, moving average and seasonal moving average parameters with 20·942 mean absolute percentage error (MAPE). Conclusions The final model showed that time series analysis of ARIMA models was useful for forecasting the number of work-related accidents in Iran. In addition, the forecasted number of work-related accidents for 2011 explained the stability of occurrence of these accidents in recent years, indicating a need for preventive occupational health and safety policies such as safety inspection. PMID:26119774
Daily Crude Oil Price Forecasting Using Hybridizing Wavelet and Artificial Neural Network Model
Directory of Open Access Journals (Sweden)
Ani Shabri
2014-01-01
Full Text Available A new method based on integrating discrete wavelet transform and artificial neural networks (WANN model for daily crude oil price forecasting is proposed. The discrete Mallat wavelet transform is used to decompose the crude price series into one approximation series and some details series (DS. The new series obtained by adding the effective one approximation series and DS component is then used as input into the ANN model to forecast crude oil price. The relative performance of WANN model was compared to regular ANN model for crude oil forecasting at lead times of 1 day for two main crude oil price series, West Texas Intermediate (WTI and Brent crude oil spot prices. In both cases, WANN model was found to provide more accurate crude oil prices forecasts than individual ANN model.
Liu, Zitao; Hauskrecht, Milos
2017-11-01
Building of an accurate predictive model of clinical time series for a patient is critical for understanding of the patient condition, its dynamics, and optimal patient management. Unfortunately, this process is not straightforward. First, patient-specific variations are typically large and population-based models derived or learned from many different patients are often unable to support accurate predictions for each individual patient. Moreover, time series observed for one patient at any point in time may be too short and insufficient to learn a high-quality patient-specific model just from the patient's own data. To address these problems we propose, develop and experiment with a new adaptive forecasting framework for building multivariate clinical time series models for a patient and for supporting patient-specific predictions. The framework relies on the adaptive model switching approach that at any point in time selects the most promising time series model out of the pool of many possible models, and consequently, combines advantages of the population, patient-specific and short-term individualized predictive models. We demonstrate that the adaptive model switching framework is very promising approach to support personalized time series prediction, and that it is able to outperform predictions based on pure population and patient-specific models, as well as, other patient-specific model adaptation strategies.
A New Strategy for Short-Term Load Forecasting
Directory of Open Access Journals (Sweden)
Yi Yang
2013-01-01
Full Text Available Electricity is a special energy which is hard to store, so the electricity demand forecasting remains an important problem. Accurate short-term load forecasting (STLF plays a vital role in power systems because it is the essential part of power system planning and operation, and it is also fundamental in many applications. Considering that an individual forecasting model usually cannot work very well for STLF, a hybrid model based on the seasonal ARIMA model and BP neural network is presented in this paper to improve the forecasting accuracy. Firstly the seasonal ARIMA model is adopted to forecast the electric load demand day ahead; then, by using the residual load demand series obtained in this forecasting process as the original series, the follow-up residual series is forecasted by BP neural network; finally, by summing up the forecasted residual series and the forecasted load demand series got by seasonal ARIMA model, the final load demand forecasting series is obtained. Case studies show that the new strategy is quite useful to improve the accuracy of STLF.
Complex dynamic behaviors of oriented percolation-based financial time series and Hang Seng index
International Nuclear Information System (INIS)
Niu, Hongli; Wang, Jun
2013-01-01
Highlights: • We develop a financial time series model by two-dimensional oriented percolation system. • We investigate the statistical behaviors of returns for HSI and the financial model by chaos-exploring methods. • We forecast the phase point of reconstructed phase space by RBF neural network. -- Abstract: We develop a financial price model by the two-dimensional oriented (directed) percolation system. The oriented percolation model is a directed variant of ordinary (isotropic) percolation, and it is applied to describe the fluctuations of stock prices. In this work, we assume that the price fluctuations result from the participants’ investment attitudes toward the market, and we investigate the information spreading among the traders and the corresponding effect on the price fluctuations. We study the complex dynamic behaviors of return time series of the model by using the multiaspect chaos-exploring methods. And we also explore the corresponding behaviors of the actual market index (Hang Seng Index) for comparison. Further, we introduce the radial basic function (RBF) neural network to train and forecast the phase point of reconstructed phase space
Fishery landing forecasting using EMD-based least square support vector machine models
Shabri, Ani
2015-05-01
In this paper, the novel hybrid ensemble learning paradigm integrating ensemble empirical mode decomposition (EMD) and least square support machine (LSSVM) is proposed to improve the accuracy of fishery landing forecasting. This hybrid is formulated specifically to address in modeling fishery landing, which has high nonlinear, non-stationary and seasonality time series which can hardly be properly modelled and accurately forecasted by traditional statistical models. In the hybrid model, EMD is used to decompose original data into a finite and often small number of sub-series. The each sub-series is modeled and forecasted by a LSSVM model. Finally the forecast of fishery landing is obtained by aggregating all forecasting results of sub-series. To assess the effectiveness and predictability of EMD-LSSVM, monthly fishery landing record data from East Johor of Peninsular Malaysia, have been used as a case study. The result shows that proposed model yield better forecasts than Autoregressive Integrated Moving Average (ARIMA), LSSVM and EMD-ARIMA models on several criteria..
Time-Series Prediction: Application to the Short-Term Electric Energy Demand
Troncoso Lora, Alicia; Riquelme Santos, Jesús Manuel; Riquelme Santos, José Cristóbal; Gómez Expósito, Antonio; Martínez Ramos, José Luis
2003-01-01
This paper describes a time-series prediction method based on the kNN technique. The proposed methodology is applied to the 24-hour load forecasting problem. Also, based on recorded data, an alternative model is developed by means of a conventional dynamic regression technique, where the parameters are estimated by solving a least squares problem. Finally, results obtained from the application of both techniques to the Spanish transmission system are compared in terms of maximum, average and ...
Time series modelling of global mean temperature for managerial decision-making.
Romilly, Peter
2005-07-01
Climate change has important implications for business and economic activity. Effective management of climate change impacts will depend on the availability of accurate and cost-effective forecasts. This paper uses univariate time series techniques to model the properties of a global mean temperature dataset in order to develop a parsimonious forecasting model for managerial decision-making over the short-term horizon. Although the model is estimated on global temperature data, the methodology could also be applied to temperature data at more localised levels. The statistical techniques include seasonal and non-seasonal unit root testing with and without structural breaks, as well as ARIMA and GARCH modelling. A forecasting evaluation shows that the chosen model performs well against rival models. The estimation results confirm the findings of a number of previous studies, namely that global mean temperatures increased significantly throughout the 20th century. The use of GARCH modelling also shows the presence of volatility clustering in the temperature data, and a positive association between volatility and global mean temperature.
Energy Technology Data Exchange (ETDEWEB)
Gay, C.; Estrada, F.; Conde, C. [Centro de Ciencias de la Atmosfera, Universidad Nacional Autonoma de Mexico, Mexico, D.F. (Mexico)]. E:mail: feporrua@atmosfera.unam.mx
2007-04-15
The common practice of using 30-year sub-samples of climatological data for describing past, present and future conditions has been widely applied, in many cases without considering the properties of the time series analyzed. This paper shows that this practice can lead to an inefficient use of the information contained in the data and to an inaccurate characterization of present, and especially future, climatological conditions because parameters are time and sub-sample size dependent. Furthermore, this approach can lead to the detection of spurious changes in distribution parameters. The time series analysis of observed monthly temperature in Veracruz, Mexico, is used to illustrate the fact that these techniques permit to make a better description of the mean and variability of the series, which in turn allows (depending on the class of process) to restrain uncertainty of forecasts, and therefore provides a better estimation of present and future risk of observing values outside a given coping range. Results presented in this paper show that, although a significant trend is found in the temperatures, giving possible evidence of observed climate change in the region, there is no evidence to support changes in the variability of the series and therefore there is neither observed evidence to support that monthly temperature variability will increase (or decrease) in the future. That is, if climate change is already occurring, it has manifested itself as a change-in-the-mean of these processes and has not affected other moments of their distributions (homogeneous non-stationary processes). The Magicc-Scengen, a software useful for constructing climate change scenarios, uses 20-year sub-samples to estimate future climate variability. For comparison purposes, possible future probability density functions are constructed following two different approaches: one, using solely the Magicc-Scengen output, and another one using a combination of this information and the time
Real time forecasting for an experimental oil spill in the arctic marginal ice zone
International Nuclear Information System (INIS)
Reed, M.; Aamo, O.M.
1994-01-01
The conference paper deals with the oil spill trajectory and weathering model OILMAP used to forecast spill trajectories for an experimental oil spill in the Barents Sea marginal ice zone. The model includes capabilities to enter graphically and display environmental data governing oil behavior: ice fields, tidal and background current fields, and wind time series, as well as geographical map information. Forecasts can also be updated from observations such as airplane overflights. The model performed well when wind was ''off-ice'' and speeds were relatively low (3-7 m/sec), with ice cover between 60% and 90%. Errors in forecasting the trajectory could be directly attributed to errors in the wind forecasts. Appropriate drift parameters for oil and ice were about 2.5% of the wind speed, with an Ekman veering angle of 35 o to the right. Ice sheets were typically 1 m thick. When the wind became ''on-ice'', speeds increased to about 10 m/sec, and trajectory simulations began to diverge from the observations, with observed drift parameters being 1.5% of the wind speed with a 60 o veering angle. Although, simple assumptions for the large scale movement of oil in dense ice fields appear appropriate, the importance of good wind forecasts as a basis for reliable trajectory prognoses cannot be overstated. 6 refs., 9 figs
Wet tropospheric delays forecast based on Vienna Mapping Function time series analysis
Rzepecka, Zofia; Kalita, Jakub
2016-04-01
It is well known that the dry part of the zenith tropospheric delay (ZTD) is much easier to model than the wet part (ZTW). The aim of the research is applying stochastic modeling and prediction of ZTW using time series analysis tools. Application of time series analysis enables closer understanding of ZTW behavior as well as short-term prediction of future ZTW values. The ZTW data used for the studies were obtained from the GGOS service hold by Vienna technical University. The resolution of the data is six hours. ZTW for the years 2010 -2013 were adopted for the study. The International GNSS Service (IGS) permanent stations LAMA and GOPE, located in mid-latitudes, were admitted for the investigations. Initially the seasonal part was separated and modeled using periodic signals and frequency analysis. The prominent annual and semi-annual signals were removed using sines and consines functions. The autocorrelation of the resulting signal is significant for several days (20-30 samples). The residuals of this fitting were further analyzed and modeled with ARIMA processes. For both the stations optimal ARMA processes based on several criterions were obtained. On this basis predicted ZTW values were computed for one day ahead, leaving the white process residuals. Accuracy of the prediction can be estimated at about 3 cm.
SEN, Jaydip; DATTA CHAUDHURI, Tamal
2016-01-01
Abstract. One of the challenging research problems in the domain of time series analysis and forecasting is making efficient and robust prediction of stock market prices. With rapid development and evolution of sophisticated algorithms and with the availability of extremely fast computing platforms, it has now become possible to effectively extract, store, process and analyze high volume stock market time series data. Complex algorithms for forecasting are now available for speedy execution o...
Monthly forecasting of agricultural pests in Switzerland
Hirschi, M.; Dubrovsky, M.; Spirig, C.; Samietz, J.; Calanca, P.; Weigel, A. P.; Fischer, A. M.; Rotach, M. W.
2012-04-01
Given the repercussions of pests and diseases on agricultural production, detailed forecasting tools have been developed to simulate the degree of infestation depending on actual weather conditions. The life cycle of pests is most successfully predicted if the micro-climate of the immediate environment (habitat) of the causative organisms can be simulated. Sub-seasonal pest forecasts therefore require weather information for the relevant habitats and the appropriate time scale. The pest forecasting system SOPRA (www.sopra.info) currently in operation in Switzerland relies on such detailed weather information, using hourly weather observations up to the day the forecast is issued, but only a climatology for the forecasting period. Here, we aim at improving the skill of SOPRA forecasts by transforming the weekly information provided by ECMWF monthly forecasts (MOFCs) into hourly weather series as required for the prediction of upcoming life phases of the codling moth, the major insect pest in apple orchards worldwide. Due to the probabilistic nature of operational monthly forecasts and the limited spatial and temporal resolution, their information needs to be post-processed for use in a pest model. In this study, we developed a statistical downscaling approach for MOFCs that includes the following steps: (i) application of a stochastic weather generator to generate a large pool of daily weather series consistent with the climate at a specific location, (ii) a subsequent re-sampling of weather series from this pool to optimally represent the evolution of the weekly MOFC anomalies, and (iii) a final extension to hourly weather series suitable for the pest forecasting model. Results show a clear improvement in the forecast skill of occurrences of upcoming codling moth life phases when incorporating MOFCs as compared to the operational pest forecasting system. This is true both in terms of root mean squared errors and of the continuous rank probability scores of the
Statistical methods for forecasting
Abraham, Bovas
2009-01-01
The Wiley-Interscience Paperback Series consists of selected books that have been made more accessible to consumers in an effort to increase global appeal and general circulation. With these new unabridged softcover volumes, Wiley hopes to extend the lives of these works by making them available to future generations of statisticians, mathematicians, and scientists."This book, it must be said, lives up to the words on its advertising cover: ''Bridging the gap between introductory, descriptive approaches and highly advanced theoretical treatises, it provides a practical, intermediate level discussion of a variety of forecasting tools, and explains how they relate to one another, both in theory and practice.'' It does just that!"-Journal of the Royal Statistical Society"A well-written work that deals with statistical methods and models that can be used to produce short-term forecasts, this book has wide-ranging applications. It could be used in the context of a study of regression, forecasting, and time series ...
Beyond Rating Curves: Time Series Models for in-Stream Turbidity Prediction
Wang, L.; Mukundan, R.; Zion, M.; Pierson, D. C.
2012-12-01
The New York City Department of Environmental Protection (DEP) manages New York City's water supply, which is comprised of over 20 reservoirs and supplies over 1 billion gallons of water per day to more than 9 million customers. DEP's "West of Hudson" reservoirs located in the Catskill Mountains are unfiltered per a renewable filtration avoidance determination granted by the EPA. While water quality is usually pristine, high volume storm events occasionally cause the reservoirs to become highly turbid. A logical strategy for turbidity control is to temporarily remove the turbid reservoirs from service. While effective in limiting delivery of turbid water and reducing the need for in-reservoir alum flocculation, this strategy runs the risk of negatively impacting water supply reliability. Thus, it is advantageous for DEP to understand how long a particular turbidity event will affect their system. In order to understand the duration, intensity and total load of a turbidity event, predictions of future in-stream turbidity values are important. Traditionally, turbidity predictions have been carried out by applying streamflow observations/forecasts to a flow-turbidity rating curve. However, predictions from rating curves are often inaccurate due to inter- and intra-event variability in flow-turbidity relationships. Predictions can be improved by applying an autoregressive moving average (ARMA) time series model in combination with a traditional rating curve. Since 2003, DEP and the Upstate Freshwater Institute have compiled a relatively consistent set of 15-minute turbidity observations at various locations on Esopus Creek above Ashokan Reservoir. Using daily averages of this data and streamflow observations at nearby USGS gauges, flow-turbidity rating curves were developed via linear regression. Time series analysis revealed that the linear regression residuals may be represented using an ARMA(1,2) process. Based on this information, flow-turbidity regressions with
Directory of Open Access Journals (Sweden)
Patrícia Ramos
2016-11-01
Full Text Available In this work, a cross-validation procedure is used to identify an appropriate Autoregressive Integrated Moving Average model and an appropriate state space model for a time series. A minimum size for the training set is specified. The procedure is based on one-step forecasts and uses different training sets, each containing one more observation than the previous one. All possible state space models and all ARIMA models where the orders are allowed to range reasonably are fitted considering raw data and log-transformed data with regular differencing (up to second order differences and, if the time series is seasonal, seasonal differencing (up to first order differences. The value of root mean squared error for each model is calculated averaging the one-step forecasts obtained. The model which has the lowest root mean squared error value and passes the Ljung–Box test using all of the available data with a reasonable significance level is selected among all the ARIMA and state space models considered. The procedure is exemplified in this paper with a case study of retail sales of different categories of women’s footwear from a Portuguese retailer, and its accuracy is compared with three reliable forecasting approaches. The results show that our procedure consistently forecasts more accurately than the other approaches and the improvements in the accuracy are significant.
Application of nonlinear forecasting techniques for meteorological modeling
Directory of Open Access Journals (Sweden)
V. Pérez-Muñuzuri
2000-10-01
Full Text Available A nonlinear forecasting method was used to predict the behavior of a cloud coverage time series several hours in advance. The method is based on the reconstruction of a chaotic strange attractor using four years of cloud absorption data obtained from half-hourly Meteosat infrared images from Northwestern Spain. An exhaustive nonlinear analysis of the time series was carried out to reconstruct the phase space of the underlying chaotic attractor. The forecast values are used by a non-hydrostatic meteorological model ARPS for daily weather prediction and their results compared with surface temperature measurements from a meteorological station and a vertical sounding. The effect of noise in the time series is analyzed in terms of the prediction results.Key words: Meterology and atmospheric dynamics (mesoscale meteorology; general – General (new fields
AN EVALUATION OF POINT AND DENSITY FORECASTS FOR SELECTED EU FARM GATE MILK PRICES
Directory of Open Access Journals (Sweden)
Dennis Bergmann
2018-01-01
Full Text Available Fundamental changes to the common agricultural policy (CAP have led to greater market orientation which in turn has resulted in sharply increased variability of EU farm gate milk prices and thus farmers’ income. In this market environment reliable forecasts of farm gate milk prices are extremely important as farmers can make improved decisions with regards to cash flow management and budget preparation. In addition these forecasts may be used in setting fixed priced contracts between dairy farmers and processors thus providing certainty and reducing risk. In this study both point and density forecasts from various time series models for farm gate milk prices in Germany, Ireland and for an average EU price series are evaluated using a rolling window framework. Additionally forecasts of the individual models are combined using different combination schemes. The results of the out of sample evaluation show that ARIMA type models perform well on short forecast horizons (1 to 3 month while the structural time series approach performs well on longer forecast horizons (12 month. Finally combining individual forecasts of different models significantly improves the forecast performance for all forecast horizons.
ESTIMATING RELIABILITY OF DISTURBANCES IN SATELLITE TIME SERIES DATA BASED ON STATISTICAL ANALYSIS
Directory of Open Access Journals (Sweden)
Z.-G. Zhou
2016-06-01
Full Text Available Normally, the status of land cover is inherently dynamic and changing continuously on temporal scale. However, disturbances or abnormal changes of land cover — caused by such as forest fire, flood, deforestation, and plant diseases — occur worldwide at unknown times and locations. Timely detection and characterization of these disturbances is of importance for land cover monitoring. Recently, many time-series-analysis methods have been developed for near real-time or online disturbance detection, using satellite image time series. However, the detection results were only labelled with “Change/ No change” by most of the present methods, while few methods focus on estimating reliability (or confidence level of the detected disturbances in image time series. To this end, this paper propose a statistical analysis method for estimating reliability of disturbances in new available remote sensing image time series, through analysis of full temporal information laid in time series data. The method consists of three main steps. (1 Segmenting and modelling of historical time series data based on Breaks for Additive Seasonal and Trend (BFAST. (2 Forecasting and detecting disturbances in new time series data. (3 Estimating reliability of each detected disturbance using statistical analysis based on Confidence Interval (CI and Confidence Levels (CL. The method was validated by estimating reliability of disturbance regions caused by a recent severe flooding occurred around the border of Russia and China. Results demonstrated that the method can estimate reliability of disturbances detected in satellite image with estimation error less than 5% and overall accuracy up to 90%.
Garcin, Matthieu
2017-10-01
Hurst exponents depict the long memory of a time series. For human-dependent phenomena, as in finance, this feature may vary in the time. It justifies modelling dynamics by multifractional Brownian motions, which are consistent with time-dependent Hurst exponents. We improve the existing literature on estimating time-dependent Hurst exponents by proposing a smooth estimate obtained by variational calculus. This method is very general and not restricted to the sole Hurst framework. It is globally more accurate and easier than other existing non-parametric estimation techniques. Besides, in the field of Hurst exponents, it makes it possible to make forecasts based on the estimated multifractional Brownian motion. The application to high-frequency foreign exchange markets (GBP, CHF, SEK, USD, CAD, AUD, JPY, CNY and SGD, all against EUR) shows significantly good forecasts. When the Hurst exponent is higher than 0.5, what depicts a long-memory feature, the accuracy is higher.
Time series analysis in road safety research uisng state space methods
BIJLEVELD, FD
2008-01-01
In this thesis we present a comprehensive study into novel time series models for aggregated road safety data. The models are mainly intended for analysis of indicators relevant to road safety, with a particular focus on how to measure these factors. Such developments may need to be related to or explained by external influences. It is also possible to make forecasts using the models. Relevant indicators include the number of persons killed permonth or year. These statistics are closely watch...
Freeway travel-time estimation and forecasting.
2012-09-01
This project presents a microsimulation-based framework for generating short-term forecasts of travel time on freeway corridors. The microsimulation model that is developed (GTsim), replicates freeway capacity drop and relaxation phenomena critical f...
Suhartono, Lee, Muhammad Hisyam; Prastyo, Dedy Dwi
2015-12-01
The aim of this research is to develop a calendar variation model for forecasting retail sales data with the Eid ul-Fitr effect. The proposed model is based on two methods, namely two levels ARIMAX and regression methods. Two levels ARIMAX and regression models are built by using ARIMAX for the first level and regression for the second level. Monthly men's jeans and women's trousers sales in a retail company for the period January 2002 to September 2009 are used as case study. In general, two levels of calendar variation model yields two models, namely the first model to reconstruct the sales pattern that already occurred, and the second model to forecast the effect of increasing sales due to Eid ul-Fitr that affected sales at the same and the previous months. The results show that the proposed two level calendar variation model based on ARIMAX and regression methods yields better forecast compared to the seasonal ARIMA model and Neural Networks.
Net load forecasting for high renewable energy penetration grids
International Nuclear Information System (INIS)
Kaur, Amanpreet; Nonnenmacher, Lukas; Coimbra, Carlos F.M.
2016-01-01
We discuss methods for net load forecasting and their significance for operation and management of power grids with high renewable energy penetration. Net load forecasting is an enabling technology for the integration of microgrid fleets with the macrogrid. Net load represents the load that is traded between the grids (microgrid and utility grid). It is important for resource allocation and electricity market participation at the point of common coupling between the interconnected grids. We compare two inherently different approaches: additive and integrated net load forecast models. The proposed methodologies are validated on a microgrid with 33% annual renewable energy (solar) penetration. A heuristics based solar forecasting technique is proposed, achieving skill of 24.20%. The integrated solar and load forecasting model outperforms the additive model by 10.69% and the uncertainty range for the additive model is larger than the integrated model by 2.2%. Thus, for grid applications an integrated forecast model is recommended. We find that the net load forecast errors and the solar forecasting errors are cointegrated with a common stochastic drift. This is useful for future planning and modeling because the solar energy time-series allows to infer important features of the net load time-series, such as expected variability and uncertainty. - Highlights: • Net load forecasting methods for grids with renewable energy generation are discussed. • Integrated solar and load forecasting outperforms the additive model by 10.69%. • Net load forecasting reduces the uncertainty between the interconnected grids.
Financial time series prediction using spiking neural networks.
Reid, David; Hussain, Abir Jaafar; Tawfik, Hissam
2014-01-01
In this paper a novel application of a particular type of spiking neural network, a Polychronous Spiking Network, was used for financial time series prediction. It is argued that the inherent temporal capabilities of this type of network are suited to non-stationary data such as this. The performance of the spiking neural network was benchmarked against three systems: two "traditional", rate-encoded, neural networks; a Multi-Layer Perceptron neural network and a Dynamic Ridge Polynomial neural network, and a standard Linear Predictor Coefficients model. For this comparison three non-stationary and noisy time series were used: IBM stock data; US/Euro exchange rate data, and the price of Brent crude oil. The experiments demonstrated favourable prediction results for the Spiking Neural Network in terms of Annualised Return and prediction error for 5-Step ahead predictions. These results were also supported by other relevant metrics such as Maximum Drawdown and Signal-To-Noise ratio. This work demonstrated the applicability of the Polychronous Spiking Network to financial data forecasting and this in turn indicates the potential of using such networks over traditional systems in difficult to manage non-stationary environments.
Short-term wind power forecasting: probabilistic and space-time aspects
DEFF Research Database (Denmark)
Tastu, Julija
work deals with the proposal and evaluation of new mathematical models and forecasting methods for short-term wind power forecasting, accounting for space-time dynamics based on geographically distributed information. Different forms of power predictions are considered, starting from traditional point...... into the corresponding models are analysed. As a final step, emphasis is placed on generating space-time trajectories: this calls for the prediction of joint multivariate predictive densities describing wind power generation at a number of distributed locations and for a number of successive lead times. In addition......Optimal integration of wind energy into power systems calls for high quality wind power predictions. State-of-the-art forecasting systems typically provide forecasts for every location individually, without taking into account information coming from the neighbouring territories. It is however...
Application of grafted polynomial function in forecasting cotton ...
African Journals Online (AJOL)
A study was conducted to forecast cotton production trend with the application of a grafted polynomial function in Nigeria from 1985 through 2013. Grafted models are used in econometrics to embark on economic analysis involving time series. In economic time series, the paucity of data and their availability has always ...
Real-Time Forecasting Revisited: Letting the Data Decide
Jackson Kitchen; John Kitchen
2013-01-01
Real-time GDP forecasting, also often known as “nowcasting,” produces estimates for current-quarter real GDP growth, typically based on a centered value from a set of estimates from incoming indicators. These real-time measures are usually intended to be data-based and to not be based on forecaster judgment or add factors. Even so, estimation methodologies in this research area—and prior versions of the system we use—typically have been constrained by using various “fixed” relationships, such...
GPS time series at Campi Flegrei caldera (2000-2013
Directory of Open Access Journals (Sweden)
Prospero De Martino
2014-05-01
Full Text Available The Campi Flegrei caldera is an active volcanic system associated to a high volcanic risk, and represents a well known and peculiar example of ground deformations (bradyseism, characterized by intense uplift periods, followed by subsidence phases with some episodic superimposed mini-uplifts. Ground deformation is an important volcanic precursor, and, its continuous monitoring, is one of the main tool for short time forecast of eruptive activity. This paper provides an overview of the continuous GPS monitoring of the Campi Flegrei caldera from January 2000 to July 2013, including network operations, data recording and processing, and data products. In this period the GPS time series allowed continuous and accurate tracking of ground deformation of the area. Seven main uplift episodes were detected, and during each uplift period, the recurrent horizontal displacement pattern, radial from the “caldera center”, suggests no significant change in deformation source geometry and location occurs. The complete archive of GPS time series at Campi Flegrei area is reported in the Supplementary materials. These data can be usefull for the scientific community in improving the research on Campi Flegrei caldera dynamic and hazard assessment.
Bao, Wei; Yue, Jun; Rao, Yulei
2017-01-01
The application of deep learning approaches to finance has received a great deal of attention from both investors and researchers. This study presents a novel deep learning framework where wavelet transforms (WT), stacked autoencoders (SAEs) and long-short term memory (LSTM) are combined for stock price forecasting. The SAEs for hierarchically extracted deep features is introduced into stock price forecasting for the first time. The deep learning framework comprises three stages. First, the stock price time series is decomposed by WT to eliminate noise. Second, SAEs is applied to generate deep high-level features for predicting the stock price. Third, high-level denoising features are fed into LSTM to forecast the next day's closing price. Six market indices and their corresponding index futures are chosen to examine the performance of the proposed model. Results show that the proposed model outperforms other similar models in both predictive accuracy and profitability performance.
Application of nonlinear forecasting techniques for meteorological modeling
Directory of Open Access Journals (Sweden)
V. Pérez-Muñuzuri
Full Text Available A nonlinear forecasting method was used to predict the behavior of a cloud coverage time series several hours in advance. The method is based on the reconstruction of a chaotic strange attractor using four years of cloud absorption data obtained from half-hourly Meteosat infrared images from Northwestern Spain. An exhaustive nonlinear analysis of the time series was carried out to reconstruct the phase space of the underlying chaotic attractor. The forecast values are used by a non-hydrostatic meteorological model ARPS for daily weather prediction and their results compared with surface temperature measurements from a meteorological station and a vertical sounding. The effect of noise in the time series is analyzed in terms of the prediction results.
Key words: Meterology and atmospheric dynamics (mesoscale meteorology; general – General (new fields
Aboagye-Sarfo, Patrick; Mai, Qun; Sanfilippo, Frank M; Preen, David B; Stewart, Louise M; Fatovich, Daniel M
2015-10-01
To develop multivariate vector-ARMA (VARMA) forecast models for predicting emergency department (ED) demand in Western Australia (WA) and compare them to the benchmark univariate autoregressive moving average (ARMA) and Winters' models. Seven-year monthly WA state-wide public hospital ED presentation data from 2006/07 to 2012/13 were modelled. Graphical and VARMA modelling methods were used for descriptive analysis and model fitting. The VARMA models were compared to the benchmark univariate ARMA and Winters' models to determine their accuracy to predict ED demand. The best models were evaluated by using error correction methods for accuracy. Descriptive analysis of all the dependent variables showed an increasing pattern of ED use with seasonal trends over time. The VARMA models provided a more precise and accurate forecast with smaller confidence intervals and better measures of accuracy in predicting ED demand in WA than the ARMA and Winters' method. VARMA models are a reliable forecasting method to predict ED demand for strategic planning and resource allocation. While the ARMA models are a closely competing alternative, they under-estimated future ED demand. Copyright © 2015 Elsevier Inc. All rights reserved.
A hybrid wind power forecasting model based on data mining and wavelets analysis
International Nuclear Information System (INIS)
Azimi, R.; Ghofrani, M.; Ghayekhloo, M.
2016-01-01
Highlights: • An improved version of K-means algorithm is proposed for clustering wind data. • A persistence based method is applied to select the best cluster for NN training. • A combination of DWT and HANTS methods is used to provide a deep learning for NN. • A hybrid of T.S.B K-means, DWT and HANTS and NN is developed for wind forecasting. - Abstract: Accurate forecasting of wind power plays a key role in energy balancing and wind power integration into the grid. This paper proposes a novel time-series based K-means clustering method, named T.S.B K-means, and a cluster selection algorithm to better extract features of wind time-series data. A hybrid of T.S.B K-means, discrete wavelet transform (DWT) and harmonic analysis time series (HANTS) methods, and a multilayer perceptron neural network (MLPNN) is developed for wind power forecasting. The proposed T.S.B K-means classifies data into separate groups and leads to more appropriate learning for neural networks by identifying anomalies and irregular patterns. This improves the accuracy of the forecast results. A cluster selection method is developed to determine the cluster that provides the best training for the MLPNN. This significantly accelerates the forecast process as the most appropriate portion of the data rather than the whole data is used for the NN training. The wind power data is decomposed by the Daubechies D4 wavelet transform, filtered by the HANTS, and pre-processed to provide the most appropriate inputs for the MLPNN. Time-series analysis is used to pre-process the historical wind-power generation data and structure it into input-output series. Wind power datasets with diverse characteristics, from different wind farms located in the United States, are used to evaluate the accuracy of the hybrid forecasting method through various performance measures and different experiments. A comparative analysis with well-established forecasting models shows the superior performance of the proposed
Probabilistic forecasting of wind power generation using extreme learning machine
DEFF Research Database (Denmark)
Wan, Can; Xu, Zhao; Pinson, Pierre
2014-01-01
an extreme learning machine (ELM)-based probabilistic forecasting method for wind power generation. To account for the uncertainties in the forecasting results, several bootstrapmethods have been compared for modeling the regression uncertainty, based on which the pairs bootstrap method is identified......Accurate and reliable forecast of wind power is essential to power system operation and control. However, due to the nonstationarity of wind power series, traditional point forecasting can hardly be accurate, leading to increased uncertainties and risks for system operation. This paper proposes...... with the best performance. Consequently, a new method for prediction intervals formulation based on theELMand the pairs bootstrap is developed.Wind power forecasting has been conducted in different seasons using the proposed approach with the historical wind power time series as the inputs alone. The results...
Short-term load forecasting of power system
Xu, Xiaobin
2017-05-01
In order to ensure the scientific nature of optimization about power system, it is necessary to improve the load forecasting accuracy. Power system load forecasting is based on accurate statistical data and survey data, starting from the history and current situation of electricity consumption, with a scientific method to predict the future development trend of power load and change the law of science. Short-term load forecasting is the basis of power system operation and analysis, which is of great significance to unit combination, economic dispatch and safety check. Therefore, the load forecasting of the power system is explained in detail in this paper. First, we use the data from 2012 to 2014 to establish the partial least squares model to regression analysis the relationship between daily maximum load, daily minimum load, daily average load and each meteorological factor, and select the highest peak by observing the regression coefficient histogram Day maximum temperature, daily minimum temperature and daily average temperature as the meteorological factors to improve the accuracy of load forecasting indicators. Secondly, in the case of uncertain climate impact, we use the time series model to predict the load data for 2015, respectively, the 2009-2014 load data were sorted out, through the previous six years of the data to forecast the data for this time in 2015. The criterion for the accuracy of the prediction is the average of the standard deviations for the prediction results and average load for the previous six years. Finally, considering the climate effect, we use the BP neural network model to predict the data in 2015, and optimize the forecast results on the basis of the time series model.
Forecasting Alcohol Consumption in the Czech Republic
Directory of Open Access Journals (Sweden)
Tereza Slováčková
2016-01-01
Full Text Available The paper deals with a forecast of developments in alcohol consumption based on current alcohol consumption per capita (expressed in litres of pure alcohol, and time series extrapolations. Alcohol consumption is to be considered from the vantage point of knowing the specifics of the product and the consequences of its excessive consumption. The predictive methodology makes use of the Box‑Jenkins method; the ARIMA model, taking into account the autocorrelation and partial autocorrelation process, which is a prerequisite for the successful identification of a time series model; model parameter estimation; appropriate transformations of time series; determining the order of differentiation and subsequent verification of the model. The chosen methodology for future trends in alcohol consumptions is a prerequisite for the proposed optional measures to control alcohol consumption in the Czech Republic. Due to the long term nature of the process to draw up and implement alcohol consumption regulation measures, the forecast covers the forthcoming 10 years.
Research on light rail electric load forecasting based on ARMA model
Huang, Yifan
2018-04-01
The article compares a variety of time series models and combines the characteristics of power load forecasting. Then, a light load forecasting model based on ARMA model is established. Based on this model, a light rail system is forecasted. The prediction results show that the accuracy of the model prediction is high.
Using forecast information for storm ride-through control
DEFF Research Database (Denmark)
Barahona Garzón, Braulio; Trombe, Pierre-Julien; Vincent, Claire Louise
2013-01-01
Using probabilistic forecast information in control algorithms can improve the performance of wind farms during periods of extreme winds. This work presents a wind farm supervisor control concept that uses probabilistic forecast information to ride-through a storm with softer ramps of power. Wind...... speed forecasts are generated with a statistical approach (i.e. time series models). The supervisor control is based on a set of logical rules that consider point forecasts and predictive densities to ramp-down the power of the wind farm before the storm hits. The potential of this supervisor control...
Automation of energy demand forecasting
Siddique, Sanzad
Automation of energy demand forecasting saves time and effort by searching automatically for an appropriate model in a candidate model space without manual intervention. This thesis introduces a search-based approach that improves the performance of the model searching process for econometrics models. Further improvements in the accuracy of the energy demand forecasting are achieved by integrating nonlinear transformations within the models. This thesis introduces machine learning techniques that are capable of modeling such nonlinearity. Algorithms for learning domain knowledge from time series data using the machine learning methods are also presented. The novel search based approach and the machine learning models are tested with synthetic data as well as with natural gas and electricity demand signals. Experimental results show that the model searching technique is capable of finding an appropriate forecasting model. Further experimental results demonstrate an improved forecasting accuracy achieved by using the novel machine learning techniques introduced in this thesis. This thesis presents an analysis of how the machine learning techniques learn domain knowledge. The learned domain knowledge is used to improve the forecast accuracy.
On statistical inference in time series analysis of the evolution of road safety.
Commandeur, Jacques J F; Bijleveld, Frits D; Bergel-Hayat, Ruth; Antoniou, Constantinos; Yannis, George; Papadimitriou, Eleonora
2013-11-01
Data collected for building a road safety observatory usually include observations made sequentially through time. Examples of such data, called time series data, include annual (or monthly) number of road traffic accidents, traffic fatalities or vehicle kilometers driven in a country, as well as the corresponding values of safety performance indicators (e.g., data on speeding, seat belt use, alcohol use, etc.). Some commonly used statistical techniques imply assumptions that are often violated by the special properties of time series data, namely serial dependency among disturbances associated with the observations. The first objective of this paper is to demonstrate the impact of such violations to the applicability of standard methods of statistical inference, which leads to an under or overestimation of the standard error and consequently may produce erroneous inferences. Moreover, having established the adverse consequences of ignoring serial dependency issues, the paper aims to describe rigorous statistical techniques used to overcome them. In particular, appropriate time series analysis techniques of varying complexity are employed to describe the development over time, relating the accident-occurrences to explanatory factors such as exposure measures or safety performance indicators, and forecasting the development into the near future. Traditional regression models (whether they are linear, generalized linear or nonlinear) are shown not to naturally capture the inherent dependencies in time series data. Dedicated time series analysis techniques, such as the ARMA-type and DRAG approaches are discussed next, followed by structural time series models, which are a subclass of state space methods. The paper concludes with general recommendations and practice guidelines for the use of time series models in road safety research. Copyright © 2012 Elsevier Ltd. All rights reserved.
Improving Artificial Neural Network Forecasts with Kalman Filtering ...
African Journals Online (AJOL)
In this paper, we examine the use of the artificial neural network method as a forecasting technique in financial time series and the application of a Kalman filter algorithm to improve the accuracy of the model. Forecasting accuracy criteria are used to compare the two models over different set of data from different companies ...
Beating the random walk: a performance assessment of long-term interest rate forecasts
den Butter, F.A.G.; Jansen, P.W.
2013-01-01
This article assesses the performance of a number of long-term interest rate forecast approaches, namely time series models, structural economic models, expert forecasts and combinations thereof. The predictive performance of these approaches is compared using outside sample forecast errors, where a
Kurniati, Devi; Hoyyi, Abdul; Widiharih, Tatik
2018-05-01
Time series data is a series of data taken or measured based on observations at the same time interval. Time series data analysis is used to perform data analysis considering the effect of time. The purpose of time series analysis is to know the characteristics and patterns of a data and predict a data value in some future period based on data in the past. One of the forecasting methods used for time series data is the state space model. This study discusses the modeling and forecasting of electric energy consumption using the state space model for univariate data. The modeling stage is began with optimal Autoregressive (AR) order selection, determination of state vector through canonical correlation analysis, estimation of parameter, and forecasting. The result of this research shows that modeling of electric energy consumption using state space model of order 4 with Mean Absolute Percentage Error (MAPE) value 3.655%, so the model is very good forecasting category.
Ensemble seasonal forecast of extreme water inflow into a large reservoir
Directory of Open Access Journals (Sweden)
A. N. Gelfan
2015-06-01
Full Text Available An approach to seasonal ensemble forecast of unregulated water inflow into a large reservoir was developed. The approach is founded on a physically-based semi-distributed hydrological model ECOMAG driven by Monte-Carlo generated ensembles of weather scenarios for a specified lead-time of the forecast (3 months ahead in this study. Case study was carried out for the Cheboksary reservoir (catchment area is 374 000 km2 located on the middle Volga River. Initial watershed conditions on the forecast date (1 March for spring freshet and 1 June for summer low-water period were simulated by the hydrological model forced by daily meteorological observations several months prior to the forecast date. A spatially distributed stochastic weather generator was used to produce time-series of daily weather scenarios for the forecast lead-time. Ensemble of daily water inflow into the reservoir was obtained by driving the ECOMAG model with the generated weather time-series. The proposed ensemble forecast technique was verified on the basis of the hindcast simulations for 29 spring and summer seasons beginning from 1982 (the year of the reservoir filling to capacity to 2010. The verification criteria were used in order to evaluate an ability of the proposed technique to forecast freshet/low-water events of the pre-assigned severity categories.
Bao, Wei; Rao, Yulei
2017-01-01
The application of deep learning approaches to finance has received a great deal of attention from both investors and researchers. This study presents a novel deep learning framework where wavelet transforms (WT), stacked autoencoders (SAEs) and long-short term memory (LSTM) are combined for stock price forecasting. The SAEs for hierarchically extracted deep features is introduced into stock price forecasting for the first time. The deep learning framework comprises three stages. First, the stock price time series is decomposed by WT to eliminate noise. Second, SAEs is applied to generate deep high-level features for predicting the stock price. Third, high-level denoising features are fed into LSTM to forecast the next day’s closing price. Six market indices and their corresponding index futures are chosen to examine the performance of the proposed model. Results show that the proposed model outperforms other similar models in both predictive accuracy and profitability performance. PMID:28708865
Directory of Open Access Journals (Sweden)
Wei Bao
Full Text Available The application of deep learning approaches to finance has received a great deal of attention from both investors and researchers. This study presents a novel deep learning framework where wavelet transforms (WT, stacked autoencoders (SAEs and long-short term memory (LSTM are combined for stock price forecasting. The SAEs for hierarchically extracted deep features is introduced into stock price forecasting for the first time. The deep learning framework comprises three stages. First, the stock price time series is decomposed by WT to eliminate noise. Second, SAEs is applied to generate deep high-level features for predicting the stock price. Third, high-level denoising features are fed into LSTM to forecast the next day's closing price. Six market indices and their corresponding index futures are chosen to examine the performance of the proposed model. Results show that the proposed model outperforms other similar models in both predictive accuracy and profitability performance.
Forecasting Macroeconomic Labour Market Flows
DEFF Research Database (Denmark)
Wilke, Ralf
2017-01-01
Forecasting labour market flows is important for budgeting and decision-making in government departments and public administration. Macroeconomic forecasts are normally obtained from time series data. In this article, we follow another approach that uses individual-level statistical analysis...... to predict the number of exits out of unemployment insurance claims. We present a comparative study of econometric, actuarial and statistical methodologies that base on different data structures. The results with records of the German unemployment insurance suggest that prediction based on individual-level...
Acute ischaemic stroke prediction from physiological time series patterns
Directory of Open Access Journals (Sweden)
Qing Zhang,
2013-05-01
Full Text Available BackgroundStroke is one of the major diseases with human mortality. Recent clinical research has indicated that early changes in common physiological variables represent a potential therapeutic target, thus the manipulation of these variables may eventually yield an effective way to optimise stroke recovery.AimsWe examined correlations between physiological parameters of patients during the first 48 hours after a stroke, and their stroke outcomes after 3 months. We wanted to discover physiological determinants that could be used to improve health outcomes by supporting the medical decisions that need to be made early on a patient’s stroke experience.Method We applied regression-based machine learning techniques to build a prediction algorithm that can forecast 3-month outcomes from initial physiological time series data during the first 48 hours after stroke. In our method, not only did we use statistical characteristics as traditional prediction features, but also we adopted trend patterns of time series data as new key features.ResultsWe tested our prediction method on a real physiological data set of stroke patients. The experiment results revealed an average high precision rate: 90%. We also tested prediction methods only considering statistical characteristics of physiological data, and concluded an average precision rate: 71%.ConclusionWe demonstrated that using trend pattern features in prediction methods improved the accuracy of stroke outcome prediction. Therefore, trend patterns of physiological time series data have an important role in the early treatment of patients with acute ischaemic stroke.
Compressive spatio-temporal forecasting of meteorological quantities and photovoltaic power
Tascikaraoglu, A.; Sanandaji, B.M.; Chicco, G.; Cocina, V.; Spertino, F.; Erdinç, O.; Paterakis, N.G.; Catalaõ, J.P.S.
2016-01-01
This paper presents a solar power forecasting scheme, which uses spatial and temporal time series data along with a photovoltaic (PV) power conversion model. The PV conversion model uses the forecast of three different variables, namely, irradiance on the tilted plane, ambient temperature, and wind
International Nuclear Information System (INIS)
Jin, Cheng Hao; Pok, Gouchol; Lee, Yongmi; Park, Hyun-Woo; Kim, Kwang Deuk; Yun, Unil; Ryu, Keun Ho
2015-01-01
Highlights: • A novel pattern sequence-based direct time series forecasting method was proposed. • Due to the use of SOM’s topology preserving property, only SOM can be applied. • SCPSNSP only deals with the cluster patterns not each specific time series value. • SCPSNSP performs better than recently developed forecasting algorithms. - Abstract: In this paper, we propose a new day-ahead direct time series forecasting method for competitive electricity markets based on clustering and next symbol prediction. In the clustering step, pattern sequence and their topology relations are obtained from self organizing map time series clustering. In the next symbol prediction step, with each cluster label in the pattern sequence represented as a pair of its topologically identical coordinates, artificial neural network is used to predict the topological coordinates of next day by training the relationship between previous daily pattern sequence and its next day pattern. According to the obtained topology relations, the nearest nonzero hits pattern is assigned to next day so that the whole time series values can be directly forecasted from the assigned cluster pattern. The proposed method was evaluated on Spanish, Australian and New York electricity markets and compared with PSF and some of the most recently published forecasting methods. Experimental results show that the proposed method outperforms the best forecasting methods at least 3.64%
A morphological perceptron with gradient-based learning for Brazilian stock market forecasting.
Araújo, Ricardo de A
2012-04-01
Several linear and non-linear techniques have been proposed to solve the stock market forecasting problem. However, a limitation arises from all these techniques and is known as the random walk dilemma (RWD). In this scenario, forecasts generated by arbitrary models have a characteristic one step ahead delay with respect to the time series values, so that, there is a time phase distortion in stock market phenomena reconstruction. In this paper, we propose a suitable model inspired by concepts in mathematical morphology (MM) and lattice theory (LT). This model is generically called the increasing morphological perceptron (IMP). Also, we present a gradient steepest descent method to design the proposed IMP based on ideas from the back-propagation (BP) algorithm and using a systematic approach to overcome the problem of non-differentiability of morphological operations. Into the learning process we have included a procedure to overcome the RWD, which is an automatic correction step that is geared toward eliminating time phase distortions that occur in stock market phenomena. Furthermore, an experimental analysis is conducted with the IMP using four complex non-linear problems of time series forecasting from the Brazilian stock market. Additionally, two natural phenomena time series are used to assess forecasting performance of the proposed IMP with other non financial time series. At the end, the obtained results are discussed and compared to results found using models recently proposed in the literature. Copyright © 2011 Elsevier Ltd. All rights reserved.
Web-Based Real Time Earthquake Forecasting and Personal Risk Management
Rundle, J. B.; Holliday, J. R.; Graves, W. R.; Turcotte, D. L.; Donnellan, A.
2012-12-01
Earthquake forecasts have been computed by a variety of countries and economies world-wide for over two decades. For the most part, forecasts have been computed for insurance, reinsurance and underwriters of catastrophe bonds. One example is the Working Group on California Earthquake Probabilities that has been responsible for the official California earthquake forecast since 1988. However, in a time of increasingly severe global financial constraints, we are now moving inexorably towards personal risk management, wherein mitigating risk is becoming the responsibility of individual members of the public. Under these circumstances, open access to a variety of web-based tools, utilities and information is a necessity. Here we describe a web-based system that has been operational since 2009 at www.openhazards.com and www.quakesim.org. Models for earthquake physics and forecasting require input data, along with model parameters. The models we consider are the Natural Time Weibull (NTW) model for regional earthquake forecasting, together with models for activation and quiescence. These models use small earthquakes ('seismicity-based models") to forecast the occurrence of large earthquakes, either through varying rates of small earthquake activity, or via an accumulation of this activity over time. These approaches use data-mining algorithms combined with the ANSS earthquake catalog. The basic idea is to compute large earthquake probabilities using the number of small earthquakes that have occurred in a region since the last large earthquake. Each of these approaches has computational challenges associated with computing forecast information in real time. Using 25 years of data from the ANSS California-Nevada catalog of earthquakes, we show that real-time forecasting is possible at a grid scale of 0.1o. We have analyzed the performance of these models using Reliability/Attributes and standard Receiver Operating Characteristic (ROC) tests. We show how the Reliability and
Directory of Open Access Journals (Sweden)
T. V. O. Fabson
2011-11-01
Full Text Available Bullwhip (or whiplash effect is an observed phenomenon in forecast driven distribution channeland careful management of these effects is of great importance to managers of supply chain.Bullwhip effect refers to situations where orders to the suppliers tend to have larger variance thansales to the buyer (demand distortion and the distortion increases as we move up the supply chain.Due to the fact that demand of customer for product is unstable, business managers must forecast inorder to properly position inventory and other resources. Forecasts are statistically based and in mostcases, are not very accurate. The existence of forecast errors made it necessary for organizations tooften carry an inventory buffer called “safety stock”. Moving up the supply chain from the end userscustomers to raw materials supplier there is a lot of variation in demand that can be observed, whichcall for greater need for safety stock.This study compares the efficacy of simulation and Time Series model in quantifying the bullwhipeffects in supply chain management.
Directory of Open Access Journals (Sweden)
Ping Jiang
2015-01-01
Full Text Available Wind speed/power has received increasing attention around the earth due to its renewable nature as well as environmental friendliness. With the global installed wind power capacity rapidly increasing, wind industry is growing into a large-scale business. Reliable short-term wind speed forecasts play a practical and crucial role in wind energy conversion systems, such as the dynamic control of wind turbines and power system scheduling. In this paper, an intelligent hybrid model for short-term wind speed prediction is examined; the model is based on cross correlation (CC analysis and a support vector regression (SVR model that is coupled with brainstorm optimization (BSO and cuckoo search (CS algorithms, which are successfully utilized for parameter determination. The proposed hybrid models were used to forecast short-term wind speeds collected from four wind turbines located on a wind farm in China. The forecasting results demonstrate that the intelligent hybrid models outperform single models for short-term wind speed forecasting, which mainly results from the superiority of BSO and CS for parameter optimization.
Johansson, Michael A.; Reich, Nicholas G.; Hota, Aditi; Brownstein, John S.; Santillana, Mauricio
2016-01-01
Dengue viruses, which infect millions of people per year worldwide, cause large epidemics that strain healthcare systems. Despite diverse efforts to develop forecasting tools including autoregressive time series, climate-driven statistical, and mechanistic biological models, little work has been done to understand the contribution of different components to improved prediction. We developed a framework to assess and compare dengue forecasts produced from different types of models and evaluate...
Machine learning based switching model for electricity load forecasting
Energy Technology Data Exchange (ETDEWEB)
Fan, Shu; Lee, Wei-Jen [Energy Systems Research Center, The University of Texas at Arlington, 416 S. College Street, Arlington, TX 76019 (United States); Chen, Luonan [Department of Electronics, Information and Communication Engineering, Osaka Sangyo University, 3-1-1 Nakagaito, Daito, Osaka 574-0013 (Japan)
2008-06-15
In deregulated power markets, forecasting electricity loads is one of the most essential tasks for system planning, operation and decision making. Based on an integration of two machine learning techniques: Bayesian clustering by dynamics (BCD) and support vector regression (SVR), this paper proposes a novel forecasting model for day ahead electricity load forecasting. The proposed model adopts an integrated architecture to handle the non-stationarity of time series. Firstly, a BCD classifier is applied to cluster the input data set into several subsets by the dynamics of the time series in an unsupervised manner. Then, groups of SVRs are used to fit the training data of each subset in a supervised way. The effectiveness of the proposed model is demonstrated with actual data taken from the New York ISO and the Western Farmers Electric Cooperative in Oklahoma. (author)
Machine learning based switching model for electricity load forecasting
Energy Technology Data Exchange (ETDEWEB)
Fan Shu [Energy Systems Research Center, University of Texas at Arlington, 416 S. College Street, Arlington, TX 76019 (United States); Chen Luonan [Department of Electronics, Information and Communication Engineering, Osaka Sangyo University, 3-1-1 Nakagaito, Daito, Osaka 574-0013 (Japan); Lee, Weijen [Energy Systems Research Center, University of Texas at Arlington, 416 S. College Street, Arlington, TX 76019 (United States)], E-mail: wlee@uta.edu
2008-06-15
In deregulated power markets, forecasting electricity loads is one of the most essential tasks for system planning, operation and decision making. Based on an integration of two machine learning techniques: Bayesian clustering by dynamics (BCD) and support vector regression (SVR), this paper proposes a novel forecasting model for day ahead electricity load forecasting. The proposed model adopts an integrated architecture to handle the non-stationarity of time series. Firstly, a BCD classifier is applied to cluster the input data set into several subsets by the dynamics of the time series in an unsupervised manner. Then, groups of SVRs are used to fit the training data of each subset in a supervised way. The effectiveness of the proposed model is demonstrated with actual data taken from the New York ISO and the Western Farmers Electric Cooperative in Oklahoma.
Machine learning based switching model for electricity load forecasting
International Nuclear Information System (INIS)
Fan Shu; Chen Luonan; Lee, Weijen
2008-01-01
In deregulated power markets, forecasting electricity loads is one of the most essential tasks for system planning, operation and decision making. Based on an integration of two machine learning techniques: Bayesian clustering by dynamics (BCD) and support vector regression (SVR), this paper proposes a novel forecasting model for day ahead electricity load forecasting. The proposed model adopts an integrated architecture to handle the non-stationarity of time series. Firstly, a BCD classifier is applied to cluster the input data set into several subsets by the dynamics of the time series in an unsupervised manner. Then, groups of SVRs are used to fit the training data of each subset in a supervised way. The effectiveness of the proposed model is demonstrated with actual data taken from the New York ISO and the Western Farmers Electric Cooperative in Oklahoma
GPS Position Time Series @ JPL
Owen, Susan; Moore, Angelyn; Kedar, Sharon; Liu, Zhen; Webb, Frank; Heflin, Mike; Desai, Shailen
2013-01-01
Different flavors of GPS time series analysis at JPL - Use same GPS Precise Point Positioning Analysis raw time series - Variations in time series analysis/post-processing driven by different users. center dot JPL Global Time Series/Velocities - researchers studying reference frame, combining with VLBI/SLR/DORIS center dot JPL/SOPAC Combined Time Series/Velocities - crustal deformation for tectonic, volcanic, ground water studies center dot ARIA Time Series/Coseismic Data Products - Hazard monitoring and response focused center dot ARIA data system designed to integrate GPS and InSAR - GPS tropospheric delay used for correcting InSAR - Caltech's GIANT time series analysis uses GPS to correct orbital errors in InSAR - Zhen Liu's talking tomorrow on InSAR Time Series analysis
New Approach To Hour-By-Hour Weather Forecast
Liao, Q. Q.; Wang, B.
2017-12-01
Fine hourly forecast in single station weather forecast is required in many human production and life application situations. Most previous MOS (Model Output Statistics) which used a linear regression model are hard to solve nonlinear natures of the weather prediction and forecast accuracy has not been sufficient at high temporal resolution. This study is to predict the future meteorological elements including temperature, precipitation, relative humidity and wind speed in a local region over a relatively short period of time at hourly level. By means of hour-to-hour NWP (Numeral Weather Prediction)meteorological field from Forcastio (https://darksky.net/dev/docs/forecast) and real-time instrumental observation including 29 stations in Yunnan and 3 stations in Tianjin of China from June to October 2016, predictions are made of the 24-hour hour-by-hour ahead. This study presents an ensemble approach to combine the information of instrumental observation itself and NWP. Use autoregressive-moving-average (ARMA) model to predict future values of the observation time series. Put newest NWP products into the equations derived from the multiple linear regression MOS technique. Handle residual series of MOS outputs with autoregressive (AR) model for the linear property presented in time series. Due to the complexity of non-linear property of atmospheric flow, support vector machine (SVM) is also introduced . Therefore basic data quality control and cross validation makes it able to optimize the model function parameters , and do 24 hours ahead residual reduction with AR/SVM model. Results show that AR model technique is better than corresponding multi-variant MOS regression method especially at the early 4 hours when the predictor is temperature. MOS-AR combined model which is comparable to MOS-SVM model outperform than MOS. Both of their root mean square error and correlation coefficients for 2 m temperature are reduced to 1.6 degree Celsius and 0.91 respectively. The
Kang, Daiwen; Mathur, Rohit; Trivikrama Rao, S.
2010-06-01
The National Air Quality Forecast Capacity (NAQFC) system, which links NOAA's North American Mesoscale (NAM) meteorological model with EPA's Community Multiscale Air Quality (CMAQ) model, provided operational ozone (O 3) and experimental fine particular matter (PM 2.5) forecasts over the continental United States (CONUS) during 2008. This paper describes the implementation of a real-time Kalman Filter (KF) bias-adjustment technique to improve the accuracy of O 3 and PM 2.5 forecasts at discrete monitoring locations. The operational surface-level O 3 and PM 2.5 forecasts from the NAQFC system were post-processed by the KF bias-adjusted technique using near real-time hourly O 3 and PM 2.5 observations obtained from EPA's AIRNow measurement network. The KF bias-adjusted forecasts were created daily, providing 24-h hourly bias-adjusted forecasts for O 3 and PM 2.5 at all AIRNow monitoring sites within the CONUS domain. The bias-adjustment post-processing implemented in this study requires minimal computational cost; requiring less than 10 min of CPU on a single processor Linux machine to generate 24-h hourly bias-adjusted forecasts over the entire CONUS domain. The results show that the real-time KF bias-adjusted forecasts for both O 3 and PM 2.5 have performed as well as or even better than the previous studies when the same technique was applied to the historical O 3 and PM 2.5 time series from archived AQF in earlier years. Compared to the raw forecasts, the KF forecasts displayed significant improvement in the daily maximum 8-h O 3 and daily mean PM 2.5 forecasts in terms of both discrete (i.e., reduced errors, increased correlation coefficients, and index of agreement) and categorical (increased hit rate and decreased false alarm ratio) evaluation metrics at almost all locations during the study period in 2008.
International Nuclear Information System (INIS)
Jafri, Y.Z.; Kamal, L.
2007-01-01
Various statistical techniques was used on five-year data from 1998-2002 of average humidity, rainfall, maximum and minimum temperatures, respectively. The relationships to regression analysis time series (RATS) were developed for determining the overall trend of these climate parameters on the basis of which forecast models can be corrected and modified. We computed the coefficient of determination as a measure of goodness of fit, to our polynomial regression analysis time series (PRATS). The correlation to multiple linear regression (MLR) and multiple linear regression analysis time series (MLRATS) were also developed for deciphering the interdependence of weather parameters. Spearman's rand correlation and Goldfeld-Quandt test were used to check the uniformity or non-uniformity of variances in our fit to polynomial regression (PR). The Breusch-Pagan test was applied to MLR and MLRATS, respectively which yielded homoscedasticity. We also employed Bartlett's test for homogeneity of variances on a five-year data of rainfall and humidity, respectively which showed that the variances in rainfall data were not homogenous while in case of humidity, were homogenous. Our results on regression and regression analysis time series show the best fit to prediction modeling on climatic data of Quetta, Pakistan. (author)
Quantifying and Reducing Uncertainty in Correlated Multi-Area Short-Term Load Forecasting
Energy Technology Data Exchange (ETDEWEB)
Sun, Yannan; Hou, Zhangshuan; Meng, Da; Samaan, Nader A.; Makarov, Yuri V.; Huang, Zhenyu
2016-07-17
In this study, we represent and reduce the uncertainties in short-term electric load forecasting by integrating time series analysis tools including ARIMA modeling, sequential Gaussian simulation, and principal component analysis. The approaches are mainly focusing on maintaining the inter-dependency between multiple geographically related areas. These approaches are applied onto cross-correlated load time series as well as their forecast errors. Multiple short-term prediction realizations are then generated from the reduced uncertainty ranges, which are useful for power system risk analyses.
Predicting physical time series using dynamic ridge polynomial neural networks.
Directory of Open Access Journals (Sweden)
Dhiya Al-Jumeily
Full Text Available Forecasting naturally occurring phenomena is a common problem in many domains of science, and this has been addressed and investigated by many scientists. The importance of time series prediction stems from the fact that it has wide range of applications, including control systems, engineering processes, environmental systems and economics. From the knowledge of some aspects of the previous behaviour of the system, the aim of the prediction process is to determine or predict its future behaviour. In this paper, we consider a novel application of a higher order polynomial neural network architecture called Dynamic Ridge Polynomial Neural Network that combines the properties of higher order and recurrent neural networks for the prediction of physical time series. In this study, four types of signals have been used, which are; The Lorenz attractor, mean value of the AE index, sunspot number, and heat wave temperature. The simulation results showed good improvements in terms of the signal to noise ratio in comparison to a number of higher order and feedforward neural networks in comparison to the benchmarked techniques.
Directory of Open Access Journals (Sweden)
Francisco Javier Duque-Pintor
2016-09-01
Full Text Available The occurrence of outliers in real-world phenomena is quite usual. If these anomalous data are not properly treated, unreliable models can be generated. Many approaches in the literature are focused on a posteriori detection of outliers. However, a new methodology to a priori predict the occurrence of such data is proposed here. Thus, the main goal of this work is to predict the occurrence of outliers in time series, by using, for the first time, imbalanced classification techniques. In this sense, the problem of forecasting outlying data has been transformed into a binary classification problem, in which the positive class represents the occurrence of outliers. Given that the number of outliers is much lower than the number of common values, the resultant classification problem is imbalanced. To create training and test sets, robust statistical methods have been used to detect outliers in both sets. Once the outliers have been detected, the instances of the dataset are labeled accordingly. Namely, if any of the samples composing the next instance are detected as an outlier, the label is set to one. As a study case, the methodology has been tested on electricity demand time series in the Spanish electricity market, in which most of the outliers were properly forecast.
Real-Time Analysis and Forecasting of Multisite River Flow Using a Distributed Hydrological Model
Directory of Open Access Journals (Sweden)
Mingdong Sun
2014-01-01
Full Text Available A spatial distributed hydrological forecasting system was developed to promote the analysis of river flow dynamic state in a large basin. The research presented the real-time analysis and forecasting of multisite river flow in the Nakdong River Basin using a distributed hydrological model with radar rainfall forecast data. A real-time calibration algorithm of hydrological distributed model was proposed to investigate the particular relationship between the water storage and basin discharge. Demonstrate the approach of simulating multisite river flow using a distributed hydrological model couple with real-time calibration and forecasting of multisite river flow with radar rainfall forecasts data. The hydrographs and results exhibit that calibrated flow simulations are very approximate to the flow observation at all sites and the accuracy of forecasting flow is gradually decreased with lead times extending from 1 hr to 3 hrs. The flow forecasts are lower than the flow observation which is likely caused by the low estimation of radar rainfall forecasts. The research has well demonstrated that the distributed hydrological model is readily applicable for multisite real-time river flow analysis and forecasting in a large basin.
Time series analysis time series analysis methods and applications
Rao, Tata Subba; Rao, C R
2012-01-01
The field of statistics not only affects all areas of scientific activity, but also many other matters such as public policy. It is branching rapidly into so many different subjects that a series of handbooks is the only way of comprehensively presenting the various aspects of statistical methodology, applications, and recent developments. The Handbook of Statistics is a series of self-contained reference books. Each volume is devoted to a particular topic in statistics, with Volume 30 dealing with time series. The series is addressed to the entire community of statisticians and scientists in various disciplines who use statistical methodology in their work. At the same time, special emphasis is placed on applications-oriented techniques, with the applied statistician in mind as the primary audience. Comprehensively presents the various aspects of statistical methodology Discusses a wide variety of diverse applications and recent developments Contributors are internationally renowened experts in their respect...
Uncertainty Analysis of Multi-Model Flood Forecasts
Directory of Open Access Journals (Sweden)
Erich J. Plate
2015-12-01
Full Text Available This paper demonstrates, by means of a systematic uncertainty analysis, that the use of outputs from more than one model can significantly improve conditional forecasts of discharges or water stages, provided the models are structurally different. Discharge forecasts from two models and the actual forecasted discharge are assumed to form a three-dimensional joint probability density distribution (jpdf, calibrated on long time series of data. The jpdf is decomposed into conditional probability density distributions (cpdf by means of Bayes formula, as suggested and explored by Krzysztofowicz in a series of papers. In this paper his approach is simplified to optimize conditional forecasts for any set of two forecast models. Its application is demonstrated by means of models developed in a study of flood forecasting for station Stung Treng on the middle reach of the Mekong River in South-East Asia. Four different forecast models were used and pairwise combined: forecast with no model, with persistence model, with a regression model, and with a rainfall-runoff model. Working with cpdfs requires determination of dependency among variables, for which linear regressions are required, as was done by Krzysztofowicz. His Bayesian approach based on transforming observed probability distributions of discharges and forecasts into normal distributions is also explored. Results obtained with his method for normal prior and likelihood distributions are identical to results from direct multiple regressions. Furthermore, it is shown that in the present case forecast accuracy is only marginally improved, if Weibull distributed basic data were converted into normally distributed variables.
International Nuclear Information System (INIS)
Benmouiza, Khalil; Cheknane, Ali
2013-01-01
Highlights: • An unsupervised clustering algorithm with a neural network model was explored. • The forecasting results of solar radiation time series and the comparison of their performance was simulated. • A new method was proposed combining k-means algorithm and NAR network to provide better prediction results. - Abstract: In this paper, we review our work for forecasting hourly global horizontal solar radiation based on the combination of unsupervised k-means clustering algorithm and artificial neural networks (ANN). k-Means algorithm focused on extracting useful information from the data with the aim of modeling the time series behavior and find patterns of the input space by clustering the data. On the other hand, nonlinear autoregressive (NAR) neural networks are powerful computational models for modeling and forecasting nonlinear time series. Taking the advantage of both methods, a new method was proposed combining k-means algorithm and NAR network to provide better forecasting results
Highly comparative time-series analysis: the empirical structure of time series and their methods.
Fulcher, Ben D; Little, Max A; Jones, Nick S
2013-06-06
The process of collecting and organizing sets of observations represents a common theme throughout the history of science. However, despite the ubiquity of scientists measuring, recording and analysing the dynamics of different processes, an extensive organization of scientific time-series data and analysis methods has never been performed. Addressing this, annotated collections of over 35 000 real-world and model-generated time series, and over 9000 time-series analysis algorithms are analysed in this work. We introduce reduced representations of both time series, in terms of their properties measured by diverse scientific methods, and of time-series analysis methods, in terms of their behaviour on empirical time series, and use them to organize these interdisciplinary resources. This new approach to comparing across diverse scientific data and methods allows us to organize time-series datasets automatically according to their properties, retrieve alternatives to particular analysis methods developed in other scientific disciplines and automate the selection of useful methods for time-series classification and regression tasks. The broad scientific utility of these tools is demonstrated on datasets of electroencephalograms, self-affine time series, heartbeat intervals, speech signals and others, in each case contributing novel analysis techniques to the existing literature. Highly comparative techniques that compare across an interdisciplinary literature can thus be used to guide more focused research in time-series analysis for applications across the scientific disciplines.
Forecasting Volatility of USD/MUR Exchange Rate using a GARCH ...
African Journals Online (AJOL)
that both distributions may forecast quite well with a slight advantage to the. GARCH(1 ... Financial time series tend to exhibit certain characteristic features such as volatility ... Heteroscedasticity-adjusted MAE to evaluate the forecasts. Chuanga et .... the Student's-t distribution or the GED with the following probability density.
Time series ARIMA models for daily price of palm oil
Ariff, Noratiqah Mohd; Zamhawari, Nor Hashimah; Bakar, Mohd Aftar Abu
2015-02-01
Palm oil is deemed as one of the most important commodity that forms the economic backbone of Malaysia. Modeling and forecasting the daily price of palm oil is of great interest for Malaysia's economic growth. In this study, time series ARIMA models are used to fit the daily price of palm oil. The Akaike Infromation Criterion (AIC), Akaike Infromation Criterion with a correction for finite sample sizes (AICc) and Bayesian Information Criterion (BIC) are used to compare between different ARIMA models being considered. It is found that ARIMA(1,2,1) model is suitable for daily price of crude palm oil in Malaysia for the year 2010 to 2012.
Introduction to Time Series Modeling
Kitagawa, Genshiro
2010-01-01
In time series modeling, the behavior of a certain phenomenon is expressed in relation to the past values of itself and other covariates. Since many important phenomena in statistical analysis are actually time series and the identification of conditional distribution of the phenomenon is an essential part of the statistical modeling, it is very important and useful to learn fundamental methods of time series modeling. Illustrating how to build models for time series using basic methods, "Introduction to Time Series Modeling" covers numerous time series models and the various tools f
Operational hydrological forecasting in Bavaria. Part I: Forecast uncertainty
Ehret, U.; Vogelbacher, A.; Moritz, K.; Laurent, S.; Meyer, I.; Haag, I.
2009-04-01
observations and several years of archived forecasts, overall empirical error distributions termed 'overall error' were for each gauge derived for a range of relevant forecast lead times. b) The error distributions vary strongly with the hydrometeorological situation, therefore a subdivision into the hydrological cases 'low flow, 'rising flood', 'flood', flood recession' was introduced. c) For the sake of numerical compression, theoretical distributions were fitted to the empirical distributions using the method of moments. Here, the normal distribution was generally best suited. d) Further data compression was achieved by representing the distribution parameters as a function (second-order polynome) of lead time. In general, the 'overall error' obtained from the above procedure is most useful in regions where large human impact occurs and where the influence of the meteorological forecast is limited. In upstream regions however, forecast uncertainty is strongly dependent on the current predictability of the atmosphere, which is contained in the spread of an ensemble forecast. Including this dynamically in the hydrological forecast uncertainty estimation requires prior elimination of the contribution of the weather forecast to the 'overall error'. This was achieved by calculating long series of hydrometeorological forecast tests, where rainfall observations were used instead of forecasts. The resulting error distribution is termed 'model error' and can be applied on hydrological ensemble forecasts, where ensemble rainfall forecasts are used as forcing. The concept will be illustrated by examples (good and bad ones) covering a wide range of catchment sizes, hydrometeorological regimes and quality of hydrological model calibration. The methodology to combine the static and dynamic shares of uncertainty will be presented in part II of this study.
Extracting Knowledge From Time Series An Introduction to Nonlinear Empirical Modeling
Bezruchko, Boris P
2010-01-01
This book addresses the fundamental question of how to construct mathematical models for the evolution of dynamical systems from experimentally-obtained time series. It places emphasis on chaotic signals and nonlinear modeling and discusses different approaches to the forecast of future system evolution. In particular, it teaches readers how to construct difference and differential model equations depending on the amount of a priori information that is available on the system in addition to the experimental data sets. This book will benefit graduate students and researchers from all natural sciences who seek a self-contained and thorough introduction to this subject.
Long Memory of Financial Time Series and Hidden Markov Models with Time-Varying Parameters
DEFF Research Database (Denmark)
Nystrup, Peter; Madsen, Henrik; Lindström, Erik
2016-01-01
Hidden Markov models are often used to model daily returns and to infer the hidden state of financial markets. Previous studies have found that the estimated models change over time, but the implications of the time-varying behavior have not been thoroughly examined. This paper presents an adaptive...... to reproduce with a hidden Markov model. Capturing the time-varying behavior of the parameters also leads to improved one-step density forecasts. Finally, it is shown that the forecasting performance of the estimated models can be further improved using local smoothing to forecast the parameter variations....
On determining the prediction limits of mathematical models for time series
International Nuclear Information System (INIS)
Peluso, E.; Gelfusa, M.; Lungaroni, M.; Talebzadeh, S.; Gaudio, P.; Murari, A.; Contributors, JET
2016-01-01
Prediction is one of the main objectives of scientific analysis and it refers to both modelling and forecasting. The determination of the limits of predictability is an important issue of both theoretical and practical relevance. In the case of modelling time series, reached a certain level in performance in either modelling or prediction, it is often important to assess whether all the information available in the data has been exploited or whether there are still margins for improvement of the tools being developed. In this paper, an information theoretic approach is proposed to address this issue and quantify the quality of the models and/or predictions. The excellent properties of the proposed indicator have been proved with the help of a systematic series of numerical tests and a concrete example of extreme relevance for nuclear fusion.
Real-time drought forecasting system for irrigation managment
Ceppi, Alessandro; Ravazzani, Giovanni; Corbari, Chiara; Masseroni, Daniele; Meucci, Stefania; Pala, Francesca; Salerno, Raffaele; Meazza, Giuseppe; Chiesa, Marco; Mancini, Marco
2013-04-01
In recent years frequent periods of water scarcity have enhanced the need to use water more carefully, even in in European areas traditionally rich of water such as the Po Valley. In dry periods, the problem of water shortage can be enhanced by conflictual use of water such as irrigation, industrial and power production (hydroelectric and thermoelectric). Further, over the last decade the social perspective on this issue is increasing due to climate change and global warming scenarios which come out from the last IPCC Report. The increased frequency of dry periods has stimulated the improvement of irrigation and water management. In this study we show the development and implementation of the real-time drought forecasting system Pre.G.I., an Italian acronym that stands for "Hydro-Meteorological forecast for irrigation management". The system is based on ensemble prediction at long range (30 days) with hydrological simulation of water balance to forecast the soil water content in every parcel over the Consorzio Muzza basin. The studied area covers 74,000 ha in the middle of the Po Valley, near the city of Lodi. The hydrological ensemble forecasts are based on 20 meteorological members of the non-hydrostatic WRF model with 30 days as lead-time, provided by Epson Meteo Centre, while the hydrological model used to generate the soil moisture and water table simulations is the rainfall-runoff distributed FEST-WB model, developed at Politecnico di Milano. The hydrological model was validated against measurements of latent heat flux and soil moisture acquired by an eddy-covariance station. Reliability of the forecasting system and its benefits was assessed on some cases-study occurred in the recent years.
Chernobyl effects on domestic and inbound tourism in Sweden. A time series analysis
International Nuclear Information System (INIS)
Hultkrantz, L.; Olsson, C.
1997-01-01
This paper estimates the impact of the Chernobyl nuclear accident on domestic and international tourism in Sweden. From ARIMA time series forecasts, outlier search, and intervention analysis based on regional monthly accommodation data from 1978-1989, no effect on domestic tourism is found. However, there is an enduring deterrence effect on incoming tourism. The loss of gross revenue from incoming tourism because of the Chernobyl accident, is estimated to 2.5 billion SEK. 5 figs., 7 tabs., 1 appendix, 27 refs
Chernobyl effects on domestic and inbound tourism in Sweden. A time series analysis
Energy Technology Data Exchange (ETDEWEB)
Hultkrantz, L. [Department of Economics, University of Uppsala, Uppsala (Sweden); Olsson, C. [Department of Economics, Umeaa University, Umeaa (Sweden)
1997-03-01
This paper estimates the impact of the Chernobyl nuclear accident on domestic and international tourism in Sweden. From ARIMA time series forecasts, outlier search, and intervention analysis based on regional monthly accommodation data from 1978-1989, no effect on domestic tourism is found. However, there is an enduring deterrence effect on incoming tourism. The loss of gross revenue from incoming tourism because of the Chernobyl accident, is estimated to 2.5 billion SEK. 5 figs., 7 tabs., 1 appendix, 27 refs.
Directory of Open Access Journals (Sweden)
Che-Jung Chang
2013-01-01
Full Text Available The wafer-level packaging process is an important technology used in semiconductor manufacturing, and how to effectively control this manufacturing system is thus an important issue for packaging firms. One way to aid in this process is to use a forecasting tool. However, the number of observations collected in the early stages of this process is usually too few to use with traditional forecasting techniques, and thus inaccurate results are obtained. One potential solution to this problem is the use of grey system theory, with its feature of small dataset modeling. This study thus uses the AGM(1,1 grey model to solve the problem of forecasting in the pilot run stage of the packaging process. The experimental results show that the grey approach is an appropriate and effective forecasting tool for use with small datasets and that it can be applied to improve the wafer-level packaging process.
Portals for Real-Time Earthquake Data and Forecasting: Challenge and Promise (Invited)
Rundle, J. B.; Holliday, J. R.; Graves, W. R.; Feltstykket, R.; Donnellan, A.; Glasscoe, M. T.
2013-12-01
Earthquake forecasts have been computed by a variety of countries world-wide for over two decades. For the most part, forecasts have been computed for insurance, reinsurance and underwriters of catastrophe bonds. However, recent events clearly demonstrate that mitigating personal risk is becoming the responsibility of individual members of the public. Open access to a variety of web-based forecasts, tools, utilities and information is therefore required. Portals for data and forecasts present particular challenges, and require the development of both apps and the client/server architecture to deliver the basic information in real time. The basic forecast model we consider is the Natural Time Weibull (NTW) method (JBR et al., Phys. Rev. E, 86, 021106, 2012). This model uses small earthquakes (';seismicity-based models') to forecast the occurrence of large earthquakes, via data-mining algorithms combined with the ANSS earthquake catalog. This method computes large earthquake probabilities using the number of small earthquakes that have occurred in a region since the last large earthquake. Localizing these forecasts in space so that global forecasts can be computed in real time presents special algorithmic challenges, which we describe in this talk. Using 25 years of data from the ANSS California-Nevada catalog of earthquakes, we compute real-time global forecasts at a grid scale of 0.1o. We analyze and monitor the performance of these models using the standard tests, which include the Reliability/Attributes and Receiver Operating Characteristic (ROC) tests. It is clear from much of the analysis that data quality is a major limitation on the accurate computation of earthquake probabilities. We discuss the challenges of serving up these datasets over the web on web-based platforms such as those at www.quakesim.org , www.e-decider.org , and www.openhazards.com.
Space-time wind speed forecasting for improved power system dispatch
Zhu, Xinxin
2014-02-27
To support large-scale integration of wind power into electric energy systems, state-of-the-art wind speed forecasting methods should be able to provide accurate and adequate information to enable efficient, reliable, and cost-effective scheduling of wind power. Here, we incorporate space-time wind forecasts into electric power system scheduling. First, we propose a modified regime-switching, space-time wind speed forecasting model that allows the forecast regimes to vary with the dominant wind direction and with the seasons, hence avoiding a subjective choice of regimes. Then, results from the wind forecasts are incorporated into a power system economic dispatch model, the cost of which is used as a loss measure of the quality of the forecast models. This, in turn, leads to cost-effective scheduling of system-wide wind generation. Potential economic benefits arise from the system-wide generation of cost savings and from the ancillary service cost savings. We illustrate the economic benefits using a test system in the northwest region of the United States. Compared with persistence and autoregressive models, our model suggests that cost savings from integration of wind power could be on the scale of tens of millions of dollars annually in regions with high wind penetration, such as Texas and the Pacific northwest. © 2014 Sociedad de Estadística e Investigación Operativa.
Functional dynamic factor models with application to yield curve forecasting
Hays, Spencer
2012-09-01
Accurate forecasting of zero coupon bond yields for a continuum of maturities is paramount to bond portfolio management and derivative security pricing. Yet a universal model for yield curve forecasting has been elusive, and prior attempts often resulted in a trade-off between goodness of fit and consistency with economic theory. To address this, herein we propose a novel formulation which connects the dynamic factor model (DFM) framework with concepts from functional data analysis: a DFM with functional factor loading curves. This results in a model capable of forecasting functional time series. Further, in the yield curve context we show that the model retains economic interpretation. Model estimation is achieved through an expectation- maximization algorithm, where the time series parameters and factor loading curves are simultaneously estimated in a single step. Efficient computing is implemented and a data-driven smoothing parameter is nicely incorporated. We show that our model performs very well on forecasting actual yield data compared with existing approaches, especially in regard to profit-based assessment for an innovative trading exercise. We further illustrate the viability of our model to applications outside of yield forecasting.
Finkenauer, C.; Gallucci, M.; van Dijk, W.; Pollmann, M.M.H.
2007-01-01
Using extensive diary data from people taking their driver's license exam, the authors investigated the role of time in affective forecasting accuracy. Replicating existing findings, participants grossly overestimated the intensity and duration of their negative affect after failure and only
Failure and reliability prediction by support vector machines regression of time series data
International Nuclear Information System (INIS)
Chagas Moura, Marcio das; Zio, Enrico; Lins, Isis Didier; Droguett, Enrique
2011-01-01
Support Vector Machines (SVMs) are kernel-based learning methods, which have been successfully adopted for regression problems. However, their use in reliability applications has not been widely explored. In this paper, a comparative analysis is presented in order to evaluate the SVM effectiveness in forecasting time-to-failure and reliability of engineered components based on time series data. The performance on literature case studies of SVM regression is measured against other advanced learning methods such as the Radial Basis Function, the traditional MultiLayer Perceptron model, Box-Jenkins autoregressive-integrated-moving average and the Infinite Impulse Response Locally Recurrent Neural Networks. The comparison shows that in the analyzed cases, SVM outperforms or is comparable to other techniques. - Highlights: → Realistic modeling of reliability demands complex mathematical formulations. → SVM is proper when the relation input/output is unknown or very costly to be obtained. → Results indicate the potential of SVM for reliability time series prediction. → Reliability estimates support the establishment of adequate maintenance strategies.
Online updating procedures for a real-time hydrological forecasting system
International Nuclear Information System (INIS)
Kahl, B; Nachtnebel, H P
2008-01-01
Rainfall-runoff-models can explain major parts of the natural runoff pattern but never simulate the observed hydrograph exactly. Reasons for errors are various sources of uncertainties embedded in the model forecasting system. Errors are due to measurement errors, the selected time period for calibration and validation, the parametric uncertainty and the model imprecision. In on-line forecasting systems forecasted input data is used which additionally generates a major uncertainty for the hydrological forecasting system. Techniques for partially compensating these uncertainties are investigated in the recent study in a medium sized catchment in the Austrian part of the Danube basin. The catchment area is about 1000 km2. The forecasting system consists of a semi-distributed continuous rainfall-runoff model that uses quantitative precipitation and temperature forecasts. To provide adequate system states at the beginning of the forecasting period continuous simulation is required, especially in winter. In this study two online updating methods are used and combined for enhancing the runoff forecasts. The first method is used for updating the system states at the beginning of the forecasting period by changing the precipitation input. The second method is an autoregressive error model, which is used to eliminate systematic errors in the model output. In combination those two methods work together well as each method is more effective in different runoff situations.
Directory of Open Access Journals (Sweden)
Chih-Chieh Young
2015-01-01
Full Text Available Accurate prediction of water level fluctuation is important in lake management due to its significant impacts in various aspects. This study utilizes four model approaches to predict water levels in the Yuan-Yang Lake (YYL in Taiwan: a three-dimensional hydrodynamic model, an artificial neural network (ANN model (back propagation neural network, BPNN, a time series forecasting (autoregressive moving average with exogenous inputs, ARMAX model, and a combined hydrodynamic and ANN model. Particularly, the black-box ANN model and physically based hydrodynamic model are coupled to more accurately predict water level fluctuation. Hourly water level data (a total of 7296 observations was collected for model calibration (training and validation. Three statistical indicators (mean absolute error, root mean square error, and coefficient of correlation were adopted to evaluate model performances. Overall, the results demonstrate that the hydrodynamic model can satisfactorily predict hourly water level changes during the calibration stage but not for the validation stage. The ANN and ARMAX models better predict the water level than the hydrodynamic model does. Meanwhile, the results from an ANN model are superior to those by the ARMAX model in both training and validation phases. The novel proposed concept using a three-dimensional hydrodynamic model in conjunction with an ANN model has clearly shown the improved prediction accuracy for the water level fluctuation.
Economic feasibility of biogas production in swine farms using time series analysis
Directory of Open Access Journals (Sweden)
Felipe Luis Rockenbach
2016-07-01
Full Text Available ABSTRACT: This study aimed to measure the economic feasibility and the time needed to return capital invested for the installation of a swine manure treatment system, these values originated the sale of carbon credits and/or of compensation of electric energy in swine farms, using the Box-Jenkins forecast models. It was found that the use of biogas is a viable option in a large scale with machines that operate daily for 10h or more, being the return period between 70 to 80 months. Time series analysis models are important to anticipate the series under study behavior, providing the swine breeder/investor means to reduce the financial investment risk as well as helping to decrease the production costs. Moreover, this process can be seen as another source of income and enable the breeder to be self-sufficient in the continuous supply of electric energy, which is very valuable nowadays considering that breeders are now increasingly using various technologies.
A methodology for Electric Power Load Forecasting
Directory of Open Access Journals (Sweden)
Eisa Almeshaiei
2011-06-01
Full Text Available Electricity demand forecasting is a central and integral process for planning periodical operations and facility expansion in the electricity sector. Demand pattern is almost very complex due to the deregulation of energy markets. Therefore, finding an appropriate forecasting model for a specific electricity network is not an easy task. Although many forecasting methods were developed, none can be generalized for all demand patterns. Therefore, this paper presents a pragmatic methodology that can be used as a guide to construct Electric Power Load Forecasting models. This methodology is mainly based on decomposition and segmentation of the load time series. Several statistical analyses are involved to study the load features and forecasting precision such as moving average and probability plots of load noise. Real daily load data from Kuwaiti electric network are used as a case study. Some results are reported to guide forecasting future needs of this network.
Applications and Comparisons of Four Time Series Models in Epidemiological Surveillance Data
Young, Alistair A.; Li, Xiaosong
2014-01-01
Public health surveillance systems provide valuable data for reliable predication of future epidemic events. This paper describes a study that used nine types of infectious disease data collected through a national public health surveillance system in mainland China to evaluate and compare the performances of four time series methods, namely, two decomposition methods (regression and exponential smoothing), autoregressive integrated moving average (ARIMA) and support vector machine (SVM). The data obtained from 2005 to 2011 and in 2012 were used as modeling and forecasting samples, respectively. The performances were evaluated based on three metrics: mean absolute error (MAE), mean absolute percentage error (MAPE), and mean square error (MSE). The accuracy of the statistical models in forecasting future epidemic disease proved their effectiveness in epidemiological surveillance. Although the comparisons found that no single method is completely superior to the others, the present study indeed highlighted that the SVMs outperforms the ARIMA model and decomposition methods in most cases. PMID:24505382
Statistical and Machine Learning forecasting methods: Concerns and ways forward
Makridakis, Spyros; Assimakopoulos, Vassilios
2018-01-01
Machine Learning (ML) methods have been proposed in the academic literature as alternatives to statistical ones for time series forecasting. Yet, scant evidence is available about their relative performance in terms of accuracy and computational requirements. The purpose of this paper is to evaluate such performance across multiple forecasting horizons using a large subset of 1045 monthly time series used in the M3 Competition. After comparing the post-sample accuracy of popular ML methods with that of eight traditional statistical ones, we found that the former are dominated across both accuracy measures used and for all forecasting horizons examined. Moreover, we observed that their computational requirements are considerably greater than those of statistical methods. The paper discusses the results, explains why the accuracy of ML models is below that of statistical ones and proposes some possible ways forward. The empirical results found in our research stress the need for objective and unbiased ways to test the performance of forecasting methods that can be achieved through sizable and open competitions allowing meaningful comparisons and definite conclusions. PMID:29584784
Statistical and Machine Learning forecasting methods: Concerns and ways forward.
Makridakis, Spyros; Spiliotis, Evangelos; Assimakopoulos, Vassilios
2018-01-01
Machine Learning (ML) methods have been proposed in the academic literature as alternatives to statistical ones for time series forecasting. Yet, scant evidence is available about their relative performance in terms of accuracy and computational requirements. The purpose of this paper is to evaluate such performance across multiple forecasting horizons using a large subset of 1045 monthly time series used in the M3 Competition. After comparing the post-sample accuracy of popular ML methods with that of eight traditional statistical ones, we found that the former are dominated across both accuracy measures used and for all forecasting horizons examined. Moreover, we observed that their computational requirements are considerably greater than those of statistical methods. The paper discusses the results, explains why the accuracy of ML models is below that of statistical ones and proposes some possible ways forward. The empirical results found in our research stress the need for objective and unbiased ways to test the performance of forecasting methods that can be achieved through sizable and open competitions allowing meaningful comparisons and definite conclusions.
Directory of Open Access Journals (Sweden)
V. Tramutoli
1996-06-01
Full Text Available An autoregressive model was selected to describe geoelectrical time series. An objective technique was subsequently applied to analyze and discriminate values above (below an a priorifixed threshold possibly related to seismic events. A complete check of the model and the main guidelines to estimate the occurrence probability of extreme events are reported. A first application of the proposed technique is discussed through the analysis of the experimental data recorded by an automatic station located in Tito, a small town on the Apennine chain in Southern Italy. This region was hit by the November 1980 Irpinia-Basilicata earthquake and it is one of most active areas of the Mediterranean region. After a preliminary filtering procedure to reduce the influence of external parameters (i.e. the meteo-climatic effects, it was demonstrated that the geoelectrical residual time series are well described by means of a second order autoregressive model. Our findings outline a statistical methodology to evaluate the efficiency of electrical seismic precursors.
Directory of Open Access Journals (Sweden)
Xiaomin Xu
2015-11-01
Full Text Available The uncertainty and regularity of wind power generation are caused by wind resources’ intermittent and randomness. Such volatility brings severe challenges to the wind power grid. The requirements for ultrashort-term and short-term wind power forecasting with high prediction accuracy of the model used, have great significance for reducing the phenomenon of abandoned wind power , optimizing the conventional power generation plan, adjusting the maintenance schedule and developing real-time monitoring systems. Therefore, accurate forecasting of wind power generation is important in electric load forecasting. The echo state network (ESN is a new recurrent neural network composed of input, hidden layer and output layers. It can approximate well the nonlinear system and achieves great results in nonlinear chaotic time series forecasting. Besides, the ESN is simpler and less computationally demanding than the traditional neural network training, which provides more accurate training results. Aiming at addressing the disadvantages of standard ESN, this paper has made some improvements. Combined with the complementary advantages of particle swarm optimization and tabu search, the generalization of ESN is improved. To verify the validity and applicability of this method, case studies of multitime scale forecasting of wind power output are carried out to reconstruct the chaotic time series of the actual wind power generation data in a certain region to predict wind power generation. Meanwhile, the influence of seasonal factors on wind power is taken into consideration. Compared with the classical ESN and the conventional Back Propagation (BP neural network, the results verify the superiority of the proposed method.
Online short-term heat load forecasting for single family houses
DEFF Research Database (Denmark)
Bacher, Peder; Madsen, Henrik; Nielsen, Henrik Aalborg
2013-01-01
. Every hour the hourly heat load for each house the following two days is forecasted. The forecast models are adaptive linear time-series models and the climate inputs used are: ambient temperature, global radiation, and wind speed. A computationally efficient recursive least squares scheme is used......This paper presents a method for forecasting the load for heating in a single-family house. Both space and hot tap water heating are forecasted. The forecasting model is built using data from sixteen houses in Sønderborg, Denmark, combined with local climate measurements and weather forecasts...... variations in the heat load signal (predominant only for some houses), peaks presumably from showers, shifts in resident behavior, and uncertainty of the weather forecasts for longer horizons, especially for the solar radiation....
International Nuclear Information System (INIS)
Tascikaraoglu, Akin; Sanandaji, Borhan M.; Poolla, Kameshwar; Varaiya, Pravin
2016-01-01
Highlights: • We propose a spatio-temporal approach for wind speed forecasting. • The method is based on a combination of Wavelet decomposition and structured-sparse recovery. • Our analyses confirm that low-dimensional structures govern the interactions between stations. • Our method particularly shows improvements for profiles with high ramps. • We examine our approach on real data and illustrate its superiority over a set of benchmark models. - Abstract: Integration of renewable energy resources into the power grid is essential in achieving the envisioned sustainable energy future. Stochasticity and intermittency characteristics of renewable energies, however, present challenges for integrating these resources into the existing grid in a large scale. Reliable renewable energy integration is facilitated by accurate wind forecasts. In this paper, we propose a novel wind speed forecasting method which first utilizes Wavelet Transform (WT) for decomposition of the wind speed data into more stationary components and then uses a spatio-temporal model on each sub-series for incorporating both temporal and spatial information. The proposed spatio-temporal forecasting approach on each sub-series is based on the assumption that there usually exists an intrinsic low-dimensional structure between time series data in a collection of meteorological stations. Our approach is inspired by Compressive Sensing (CS) and structured-sparse recovery algorithms. Based on detailed case studies, we show that the proposed approach based on exploiting the sparsity of correlations between a large set of meteorological stations and decomposing time series for higher-accuracy forecasts considerably improve the short-term forecasts compared to the temporal and spatio-temporal benchmark methods.
Environmental noise forecasting based on support vector machine
Fu, Yumei; Zan, Xinwu; Chen, Tianyi; Xiang, Shihan
2018-01-01
As an important pollution source, the noise pollution is always the researcher's focus. Especially in recent years, the noise pollution is seriously harmful to the human beings' environment, so the research about the noise pollution is a very hot spot. Some noise monitoring technologies and monitoring systems are applied in the environmental noise test, measurement and evaluation. But, the research about the environmental noise forecasting is weak. In this paper, a real-time environmental noise monitoring system is introduced briefly. This monitoring system is working in Mianyang City, Sichuan Province. It is monitoring and collecting the environmental noise about more than 20 enterprises in this district. Based on the large amount of noise data, the noise forecasting by the Support Vector Machine (SVM) is studied in detail. Compared with the time series forecasting model and the artificial neural network forecasting model, the SVM forecasting model has some advantages such as the smaller data size, the higher precision and stability. The noise forecasting results based on the SVM can provide the important and accuracy reference to the prevention and control of the environmental noise.
Frequency domain methods applied to forecasting electricity markets
International Nuclear Information System (INIS)
Trapero, Juan R.; Pedregal, Diego J.
2009-01-01
The changes taking place in electricity markets during the last two decades have produced an increased interest in the problem of forecasting, either load demand or prices. Many forecasting methodologies are available in the literature nowadays with mixed conclusions about which method is most convenient. This paper focuses on the modeling of electricity market time series sampled hourly in order to produce short-term (1 to 24 h ahead) forecasts. The main features of the system are that (1) models are of an Unobserved Component class that allow for signal extraction of trend, diurnal, weekly and irregular components; (2) its application is automatic, in the sense that there is no need for human intervention via any sort of identification stage; (3) the models are estimated in the frequency domain; and (4) the robustness of the method makes possible its direct use on both load demand and price time series. The approach is thoroughly tested on the PJM interconnection market and the results improve on classical ARIMA models. (author)
Forecasting the Emergency Department Patients Flow.
Afilal, Mohamed; Yalaoui, Farouk; Dugardin, Frédéric; Amodeo, Lionel; Laplanche, David; Blua, Philippe
2016-07-01
Emergency department (ED) have become the patient's main point of entrance in modern hospitals causing it frequent overcrowding, thus hospital managers are increasingly paying attention to the ED in order to provide better quality service for patients. One of the key elements for a good management strategy is demand forecasting. In this case, forecasting patients flow, which will help decision makers to optimize human (doctors, nurses…) and material(beds, boxs…) resources allocation. The main interest of this research is forecasting daily attendance at an emergency department. The study was conducted on the Emergency Department of Troyes city hospital center, France, in which we propose a new practical ED patients classification that consolidate the CCMU and GEMSA categories into one category and innovative time-series based models to forecast long and short term daily attendance. The models we developed for this case study shows very good performances (up to 91,24 % for the annual Total flow forecast) and robustness to epidemic periods.
Mukhin, Dmitry; Gavrilov, Andrey; Loskutov, Evgeny; Feigin, Alexander
2016-04-01
We suggest a method for empirical forecast of climate dynamics basing on the reconstruction of reduced dynamical models in a form of random dynamical systems [1,2] derived from observational time series. The construction of proper embedding - the set of variables determining the phase space the model works in - is no doubt the most important step in such a modeling, but this task is non-trivial due to huge dimension of time series of typical climatic fields. Actually, an appropriate expansion of observational time series is needed yielding the number of principal components considered as phase variables, which are to be efficient for the construction of low-dimensional evolution operator. We emphasize two main features the reduced models should have for capturing the main dynamical properties of the system: (i) taking into account time-lagged teleconnections in the atmosphere-ocean system and (ii) reflecting the nonlinear nature of these teleconnections. In accordance to these principles, in this report we present the methodology which includes the combination of a new way for the construction of an embedding by the spatio-temporal data expansion and nonlinear model construction on the basis of artificial neural networks. The methodology is aplied to NCEP/NCAR reanalysis data including fields of sea level pressure, geopotential height, and wind speed, covering Northern Hemisphere. Its efficiency for the interannual forecast of various climate phenomena including ENSO, PDO, NAO and strong blocking event condition over the mid latitudes, is demonstrated. Also, we investigate the ability of the models to reproduce and predict the evolution of qualitative features of the dynamics, such as spectral peaks, critical transitions and statistics of extremes. This research was supported by the Government of the Russian Federation (Agreement No. 14.Z50.31.0033 with the Institute of Applied Physics RAS) [1] Y. I. Molkov, E. M. Loskutov, D. N. Mukhin, and A. M. Feigin, "Random
Gold sales forecasting: The Box-Jenkins methodology
Directory of Open Access Journals (Sweden)
Johannes Tshepiso Tsoku
2017-02-01
Full Text Available The study employs the Box-Jenkins Methodology to forecast South African gold sales. For a resource economy like South Africa where metals and minerals account for a high proportion of GDP and export earnings, the decline in gold sales is very disturbing. Box-Jenkins time series technique was used to perform time series analysis of monthly gold sales for the period January 2000 to June 2013 with the following steps: model identification, model estimation, diagnostic checking and forecasting. Furthermore, the prediction accuracy is tested using mean absolute percentage error (MAPE. From the analysis, a seasonal ARIMA(4,1,4×(0,1,112 was found to be the “best fit model” with an MAPE value of 11% indicating that the model is fit to be used to predict or forecast future gold sales for South Africa. In addition, the forecast values show that there will be a decrease in the overall gold sales for the first six months of 2014. It is hoped that the study will help the public and private sectors to understand the gold sales or output scenario and later plan the gold mining activities in South Africa. Furthermore, it is hoped that this research paper has demonstrated the significance of Box-Jenkins technique for this area of research and that they will be applied in the future.
Treatment of Outliers via Interpolation Method with Neural Network Forecast Performances
Wahir, N. A.; Nor, M. E.; Rusiman, M. S.; Gopal, K.
2018-04-01
Outliers often lurk in many datasets, especially in real data. Such anomalous data can negatively affect statistical analyses, primarily normality, variance, and estimation aspects. Hence, handling the occurrences of outliers require special attention. Therefore, it is important to determine the suitable ways in treating outliers so as to ensure that the quality of the analyzed data is indeed high. As such, this paper discusses an alternative method to treat outliers via linear interpolation method. In fact, assuming outlier as a missing value in the dataset allows the application of the interpolation method to interpolate the outliers thus, enabling the comparison of data series using forecast accuracy before and after outlier treatment. With that, the monthly time series of Malaysian tourist arrivals from January 1998 until December 2015 had been used to interpolate the new series. The results indicated that the linear interpolation method, which was comprised of improved time series data, displayed better results, when compared to the original time series data in forecasting from both Box-Jenkins and neural network approaches.
Identifying secondary series for stepwise common singular spectrum ...
African Journals Online (AJOL)
Abstract. Common singular spectrum analysis is a technique which can be used to forecast a pri- mary time series by using the information from a secondary series. Not all secondary series, however, provide useful information. A first contribution in this paper is to point out the properties which a secondary series should ...
Short-term electric load forecasting using computational intelligence methods
Jurado, Sergio; Peralta, J.; Nebot, Àngela; Mugica, Francisco; Cortez, Paulo
2013-01-01
Accurate time series forecasting is a key issue to support individual and organizational decision making. In this paper, we introduce several methods for short-term electric load forecasting. All the presented methods stem from computational intelligence techniques: Random Forest, Nonlinear Autoregressive Neural Networks, Evolutionary Support Vector Machines and Fuzzy Inductive Reasoning. The performance of the suggested methods is experimentally justified with several experiments carried out...
Research on PM2.5 time series characteristics based on data mining technology
Zhao, Lifang; Jia, Jin
2018-02-01
With the development of data mining technology and the establishment of environmental air quality database, it is necessary to discover the potential correlations and rules by digging the massive environmental air quality information and analyzing the air pollution process. In this paper, we have presented a sequential pattern mining method based on the air quality data and pattern association technology to analyze the PM2.5 time series characteristics. Utilizing the real-time monitoring data of urban air quality in China, the time series rule and variation properties of PM2.5 under different pollution levels are extracted and analyzed. The analysis results show that the time sequence features of the PM2.5 concentration is directly affected by the alteration of the pollution degree. The longest time that PM2.5 remained stable is about 24 hours. As the pollution degree gets severer, the instability time and step ascending time gradually changes from 12-24 hours to 3 hours. The presented method is helpful for the controlling and forecasting of the air quality while saving the measuring costs, which is of great significance for the government regulation and public prevention of the air pollution.
Shimada, Yutaka; Ikeguchi, Tohru; Shigehara, Takaomi
2012-10-01
In this Letter, we propose a framework to transform a complex network to a time series. The transformation from complex networks to time series is realized by the classical multidimensional scaling. Applying the transformation method to a model proposed by Watts and Strogatz [Nature (London) 393, 440 (1998)], we show that ring lattices are transformed to periodic time series, small-world networks to noisy periodic time series, and random networks to random time series. We also show that these relationships are analytically held by using the circulant-matrix theory and the perturbation theory of linear operators. The results are generalized to several high-dimensional lattices.
Vintage errors: do real-time economic data improve election forecasts?
Directory of Open Access Journals (Sweden)
Mark Andreas Kayser
2015-07-01
Full Text Available Economic performance is a key component of most election forecasts. When fitting models, however, most forecasters unwittingly assume that the actual state of the economy, a state best estimated by the multiple periodic revisions to official macroeconomic statistics, drives voter behavior. The difference in macroeconomic estimates between revised and original data vintages can be substantial, commonly over 100% (two-fold for economic growth estimates, making the choice of which data release to use important for the predictive validity of a model. We systematically compare the predictions of four forecasting models for numerous US presidential elections using real-time and vintage data. We find that newer data are not better data for election forecasting: forecasting error increases with data revisions. This result suggests that voter perceptions of economic growth are influenced more by media reports about the economy, which are based on initial economic estimates, than by the actual state of the economy.
Hybrid Intrusion Forecasting Framework for Early Warning System
Kim, Sehun; Shin, Seong-Jun; Kim, Hyunwoo; Kwon, Ki Hoon; Han, Younggoo
Recently, cyber attacks have become a serious hindrance to the stability of Internet. These attacks exploit interconnectivity of networks, propagate in an instant, and have become more sophisticated and evolutionary. Traditional Internet security systems such as firewalls, IDS and IPS are limited in terms of detecting recent cyber attacks in advance as these systems respond to Internet attacks only after the attacks inflict serious damage. In this paper, we propose a hybrid intrusion forecasting system framework for an early warning system. The proposed system utilizes three types of forecasting methods: time-series analysis, probabilistic modeling, and data mining method. By combining these methods, it is possible to take advantage of the forecasting technique of each while overcoming their drawbacks. Experimental results show that the hybrid intrusion forecasting method outperforms each of three forecasting methods.
Tourism Demand Modelling and Forecasting: A Review of Recent Research
Song, H; Li, G
2008-01-01
This paper reviews the published studies on tourism demand modelling and forecasting since 2000. One of the key findings of this review is that the methods used in analysing and forecasting the demand for tourism have been more diverse than those identified by other review articles. In addition to the most popular time-series and econometric models, a number of new techniques have emerged in the literature. However, as far as the forecasting accuracy is concerned, the study shows that there i...
A hierarchical bayesian model to quantify uncertainty of stream water temperature forecasts.
Directory of Open Access Journals (Sweden)
Guillaume Bal
Full Text Available Providing generic and cost effective modelling approaches to reconstruct and forecast freshwater temperature using predictors as air temperature and water discharge is a prerequisite to understanding ecological processes underlying the impact of water temperature and of global warming on continental aquatic ecosystems. Using air temperature as a simple linear predictor of water temperature can lead to significant bias in forecasts as it does not disentangle seasonality and long term trends in the signal. Here, we develop an alternative approach based on hierarchical Bayesian statistical time series modelling of water temperature, air temperature and water discharge using seasonal sinusoidal periodic signals and time varying means and amplitudes. Fitting and forecasting performances of this approach are compared with that of simple linear regression between water and air temperatures using i an emotive simulated example, ii application to three French coastal streams with contrasting bio-geographical conditions and sizes. The time series modelling approach better fit data and does not exhibit forecasting bias in long term trends contrary to the linear regression. This new model also allows for more accurate forecasts of water temperature than linear regression together with a fair assessment of the uncertainty around forecasting. Warming of water temperature forecast by our hierarchical Bayesian model was slower and more uncertain than that expected with the classical regression approach. These new forecasts are in a form that is readily usable in further ecological analyses and will allow weighting of outcomes from different scenarios to manage climate change impacts on freshwater wildlife.
ENSEMBLE methods to reconcile disparate national long range dispersion forecasting
Energy Technology Data Exchange (ETDEWEB)
Mikkelsen, T; Galmarini, S; Bianconi, R; French, S [eds.
2003-11-01
ENSEMBLE is a web-based decision support system for real-time exchange and evaluation of national long-range dispersion forecasts of nuclear releases with cross-boundary consequences. The system is developed with the purpose to reconcile among disparate national forecasts for long-range dispersion. ENSEMBLE addresses the problem of achieving a common coherent strategy across European national emergency management when national long-range dispersion forecasts differ from one another during an accidental atmospheric release of radioactive material. A series of new decision-making 'ENSEMBLE' procedures and Web-based software evaluation and exchange tools have been created for real-time reconciliation and harmonisation of real-time dispersion forecasts from meteorological and emergency centres across Europe during an accident. The new ENSEMBLE software tools is available to participating national emergency and meteorological forecasting centres, which may choose to integrate them directly into operational emergency information systems, or possibly use them as a basis for future system development. (au)
ENSEMBLE methods to reconcile disparate national long range dispersion forecasting
Energy Technology Data Exchange (ETDEWEB)
Mikkelsen, T.; Galmarini, S.; Bianconi, R.; French, S. (eds.)
2003-11-01
ENSEMBLE is a web-based decision support system for real-time exchange and evaluation of national long-range dispersion forecasts of nuclear releases with cross-boundary consequences. The system is developed with the purpose to reconcile among disparate national forecasts for long-range dispersion. ENSEMBLE addresses the problem of achieving a common coherent strategy across European national emergency management when national long-range dispersion forecasts differ from one another during an accidental atmospheric release of radioactive material. A series of new decision-making 'ENSEMBLE' procedures and Web-based software evaluation and exchange tools have been created for real-time reconciliation and harmonisation of real-time dispersion forecasts from meteorological and emergency centres across Europe during an accident. The new ENSEMBLE software tools is available to participating national emergency and meteorological forecasting centres, which may choose to integrate them directly into operational emergency information systems, or possibly use them as a basis for future system development. (au)
Duality between Time Series and Networks
Campanharo, Andriana S. L. O.; Sirer, M. Irmak; Malmgren, R. Dean; Ramos, Fernando M.; Amaral, Luís A. Nunes.
2011-01-01
Studying the interaction between a system's components and the temporal evolution of the system are two common ways to uncover and characterize its internal workings. Recently, several maps from a time series to a network have been proposed with the intent of using network metrics to characterize time series. Although these maps demonstrate that different time series result in networks with distinct topological properties, it remains unclear how these topological properties relate to the original time series. Here, we propose a map from a time series to a network with an approximate inverse operation, making it possible to use network statistics to characterize time series and time series statistics to characterize networks. As a proof of concept, we generate an ensemble of time series ranging from periodic to random and confirm that application of the proposed map retains much of the information encoded in the original time series (or networks) after application of the map (or its inverse). Our results suggest that network analysis can be used to distinguish different dynamic regimes in time series and, perhaps more importantly, time series analysis can provide a powerful set of tools that augment the traditional network analysis toolkit to quantify networks in new and useful ways. PMID:21858093
International Workshop on Industry Practices for Forecasting
Poggi, Jean-Michel; Brossat, Xavier
2015-01-01
The chapters in this volume stress the need for advances in theoretical understanding to go hand-in-hand with the widespread practical application of forecasting in industry. Forecasting and time series prediction have enjoyed considerable attention over the last few decades, fostered by impressive advances in observational capabilities and measurement procedures. On June 5-7, 2013, an international Workshop on Industry Practices for FORecasting was held in Paris, France, organized and supported by the OSIRIS Department of Electricité de France Research and Development Division. In keeping with tradition, both theoretical statistical results and practical contributions on this active field of statistical research and on forecasting issues in a rapidly evolving industrial environment are presented. The volume reflects the broad spectrum of the conference, including 16 articles contributed by specialists in various areas. The material compiled is broad in scope and ranges from new findings on forecasting in in...
DEFF Research Database (Denmark)
Hisdal, H.; Holmqvist, E.; Hyvärinen, V.
Awareness that emission of greenhouse gases will raise the global temperature and change the climate has led to studies trying to identify such changes in long-term climate and hydrologic time series. This report, written by the......Awareness that emission of greenhouse gases will raise the global temperature and change the climate has led to studies trying to identify such changes in long-term climate and hydrologic time series. This report, written by the...
Forecasting oil price trends using wavelets and hidden Markov models
International Nuclear Information System (INIS)
Souza e Silva, Edmundo G. de; Souza e Silva, Edmundo A. de; Legey, Luiz F.L.
2010-01-01
The crude oil price is influenced by a great number of factors, most of which interact in very complex ways. For this reason, forecasting it through a fundamentalist approach is a difficult task. An alternative is to use time series methodologies, with which the price's past behavior is conveniently analyzed, and used to predict future movements. In this paper, we investigate the usefulness of a nonlinear time series model, known as hidden Markov model (HMM), to predict future crude oil price movements. Using an HMM, we develop a forecasting methodology that consists of, basically, three steps. First, we employ wavelet analysis to remove high frequency price movements, which can be assumed as noise. Then, the HMM is used to forecast the probability distribution of the price return accumulated over the next F days. Finally, from this distribution, we infer future price trends. Our results indicate that the proposed methodology might be a useful decision support tool for agents participating in the crude oil market. (author)
Evaluating the Effectiveness of DART® Buoy Networks Based on Forecast Accuracy
Percival, Donald B.; Denbo, Donald W.; Gica, Edison; Huang, Paul Y.; Mofjeld, Harold O.; Spillane, Michael C.; Titov, Vasily V.
2018-03-01
A performance measure for a DART® tsunami buoy network has been developed. DART® buoys are used to detect tsunamis, but the full potential of the data they collect is realized through accurate forecasts of inundations caused by the tsunamis. The performance measure assesses how well the network achieves its full potential through a statistical analysis of simulated forecasts of wave amplitudes outside an impact site and a consideration of how much the forecasts are degraded in accuracy when one or more buoys are inoperative. The analysis uses simulated tsunami amplitude time series collected at each buoy from selected source segments in the Short-term Inundation Forecast for Tsunamis database and involves a set for 1000 forecasts for each buoy/segment pair at sites just offshore of selected impact communities. Random error-producing scatter in the time series is induced by uncertainties in the source location, addition of real oceanic noise, and imperfect tidal removal. Comparison with an error-free standard leads to root-mean-square errors (RMSEs) for DART® buoys located near a subduction zone. The RMSEs indicate which buoy provides the best forecast (lowest RMSE) for sections of the zone, under a warning-time constraint for the forecasts of 3 h. The analysis also shows how the forecasts are degraded (larger minimum RMSE among the remaining buoys) when one or more buoys become inoperative. The RMSEs provide a way to assess array augmentation or redesign such as moving buoys to more optimal locations. Examples are shown for buoys off the Aleutian Islands and off the West Coast of South America for impact sites at Hilo HI and along the US West Coast (Crescent City CA and Port San Luis CA, USA). A simple measure (coded green, yellow or red) of the current status of the network's ability to deliver accurate forecasts is proposed to flag the urgency of buoy repair.
Evaluating the Effectiveness of DART® Buoy Networks Based on Forecast Accuracy
Percival, Donald B.; Denbo, Donald W.; Gica, Edison; Huang, Paul Y.; Mofjeld, Harold O.; Spillane, Michael C.; Titov, Vasily V.
2018-04-01
A performance measure for a DART® tsunami buoy network has been developed. DART® buoys are used to detect tsunamis, but the full potential of the data they collect is realized through accurate forecasts of inundations caused by the tsunamis. The performance measure assesses how well the network achieves its full potential through a statistical analysis of simulated forecasts of wave amplitudes outside an impact site and a consideration of how much the forecasts are degraded in accuracy when one or more buoys are inoperative. The analysis uses simulated tsunami amplitude time series collected at each buoy from selected source segments in the Short-term Inundation Forecast for Tsunamis database and involves a set for 1000 forecasts for each buoy/segment pair at sites just offshore of selected impact communities. Random error-producing scatter in the time series is induced by uncertainties in the source location, addition of real oceanic noise, and imperfect tidal removal. Comparison with an error-free standard leads to root-mean-square errors (RMSEs) for DART® buoys located near a subduction zone. The RMSEs indicate which buoy provides the best forecast (lowest RMSE) for sections of the zone, under a warning-time constraint for the forecasts of 3 h. The analysis also shows how the forecasts are degraded (larger minimum RMSE among the remaining buoys) when one or more buoys become inoperative. The RMSEs provide a way to assess array augmentation or redesign such as moving buoys to more optimal locations. Examples are shown for buoys off the Aleutian Islands and off the West Coast of South America for impact sites at Hilo HI and along the US West Coast (Crescent City CA and Port San Luis CA, USA). A simple measure (coded green, yellow or red) of the current status of the network's ability to deliver accurate forecasts is proposed to flag the urgency of buoy repair.
Luo, Li; Luo, Le; Zhang, Xinli; He, Xiaoli
2017-07-10
Accurate forecasting of hospital outpatient visits is beneficial for the reasonable planning and allocation of healthcare resource to meet the medical demands. In terms of the multiple attributes of daily outpatient visits, such as randomness, cyclicity and trend, time series methods, ARIMA, can be a good choice for outpatient visits forecasting. On the other hand, the hospital outpatient visits are also affected by the doctors' scheduling and the effects are not pure random. Thinking about the impure specialty, this paper presents a new forecasting model that takes cyclicity and the day of the week effect into consideration. We formulate a seasonal ARIMA (SARIMA) model on a daily time series and then a single exponential smoothing (SES) model on the day of the week time series, and finally establish a combinatorial model by modifying them. The models are applied to 1 year of daily visits data of urban outpatients in two internal medicine departments of a large hospital in Chengdu, for forecasting the daily outpatient visits about 1 week ahead. The proposed model is applied to forecast the cross-sectional data for 7 consecutive days of daily outpatient visits over an 8-weeks period based on 43 weeks of observation data during 1 year. The results show that the two single traditional models and the combinatorial model are simplicity of implementation and low computational intensiveness, whilst being appropriate for short-term forecast horizons. Furthermore, the combinatorial model can capture the comprehensive features of the time series data better. Combinatorial model can achieve better prediction performance than the single model, with lower residuals variance and small mean of residual errors which needs to be optimized deeply on the next research step.
A Course in Time Series Analysis
Peña, Daniel; Tsay, Ruey S
2011-01-01
New statistical methods and future directions of research in time series A Course in Time Series Analysis demonstrates how to build time series models for univariate and multivariate time series data. It brings together material previously available only in the professional literature and presents a unified view of the most advanced procedures available for time series model building. The authors begin with basic concepts in univariate time series, providing an up-to-date presentation of ARIMA models, including the Kalman filter, outlier analysis, automatic methods for building ARIMA models, a
ENSEMBLE methods to reconcile disparate national long range dispersion forecasts
DEFF Research Database (Denmark)
Mikkelsen, Torben; Galmarini, S.; Bianconi, R.
2003-01-01
ENSEMBLE is a web-based decision support system for real-time exchange and evaluation of national long-range dispersion forecasts of nuclear releases with cross-boundary consequences. The system is developed with the purpose to reconcile among disparatenational forecasts for long-range dispersion...... emergency and meteorological forecasting centres, which may choose to integrate them directly intooperational emergency information systems, or possibly use them as a basis for future system development.......ENSEMBLE is a web-based decision support system for real-time exchange and evaluation of national long-range dispersion forecasts of nuclear releases with cross-boundary consequences. The system is developed with the purpose to reconcile among disparatenational forecasts for long-range dispersion....... ENSEMBLE addresses the problem of achieving a common coherent strategy across European national emergency management when national long-range dispersion forecasts differ from one another during an accidentalatmospheric release of radioactive material. A series of new decision-making “ENSEMBLE” procedures...
Directory of Open Access Journals (Sweden)
Yuyang Gao
2016-09-01
Full Text Available With increasing importance being attached to big data mining, analysis, and forecasting in the field of wind energy, how to select an optimization model to improve the forecasting accuracy of the wind speed time series is not only an extremely challenging problem, but also a problem of concern for economic forecasting. The artificial intelligence model is widely used in forecasting and data processing, but the individual back-propagation artificial neural network cannot always satisfy the time series forecasting needs. Thus, a hybrid forecasting approach has been proposed in this study, which consists of data preprocessing, parameter optimization and a neural network for advancing the accuracy of short-term wind speed forecasting. According to the case study, in which the data are collected from Peng Lai, a city located in China, the simulation results indicate that the hybrid forecasting method yields better predictions compared to the individual BP, which indicates that the hybrid method exhibits stronger forecasting ability.
Directory of Open Access Journals (Sweden)
Ibgtc Bowala
2017-06-01
Full Text Available With the rapid growth of financial markets, analyzers are paying more attention on predictions. Stock data are time series data, with huge amounts. Feasible solution for handling the increasing amount of data is to use a cluster for parallel processing, and Hadoop parallel computing platform is a typical representative. There are various statistical models for forecasting time series data, but accurate clusters are a pre-requirement. Clustering analysis for time series data is one of the main methods for mining time series data for many other analysis processes. However, general clustering algorithms cannot perform clustering for time series data because series data has a special structure and a high dimensionality has highly co-related values due to high noise level. A novel model for time series clustering is presented using BIRCH, based on piecewise SVD, leading to a novel dimension reduction approach. Highly co-related features are handled using SVD with a novel approach for dimensionality reduction in order to keep co-related behavior optimal and then use BIRCH for clustering. The algorithm is a novel model that can handle massive time series data. Finally, this new model is successfully applied to real stock time series data of Yahoo finance with satisfactory results.
DEFF Research Database (Denmark)
Nielsen, Joakim Refslund; Dellwik, Ebba; Hahmann, Andrea N.
2014-01-01
A method is presented for development of satellite green vegetation fraction (GVF) time series for use in the Weather Research and Forecasting (WRF) model. The GVF data is in the WRF model used to describe the temporal evolution of many land surface parameters, in addition to the evolution of veg...
Short term solar radiation forecasting: Island versus continental sites
International Nuclear Information System (INIS)
Boland, John; David, Mathieu; Lauret, Philippe
2016-01-01
Due its intermittency, the large-scale integration of solar energy into electricity grids is an issue and more specifically in an insular context. Thus, forecasting the output of solar energy is a key feature to efficiently manage the supply-demand balance. In this paper, three short term forecasting procedures are applied to island locations in order to see how they perform in situations that are potentially more volatile than continental locations. Two continental locations, one coastal and one inland are chosen for comparison. At the two time scales studied, ten minute and hourly, the island locations prove to be more difficult to forecast, as shown by larger forecast errors. It is found that the three methods, one purely statistical combining Fourier series plus linear ARMA models, one combining clear sky index models plus neural net models, and a third using a clear sky index plus ARMA, give similar forecasting results. It is also suggested that there is great potential of merging modelling approaches on different horizons. - Highlights: • Solar energy forecasting is more difficult for insular than continental sites. • Fourier series plus linear ARMA models are one forecasting method tested. • Clear sky index models plus neural net models are also tested. • Clear sky index models plus linear ARMA is also an option. • All three approaches have similar skill.
Pérez-Jordán, G.; Castro-Almazán, J. A.; Muñoz-Tuñón, C.
2018-04-01
We validate the Weather Research and Forecasting (WRF) model for precipitable water vapour (PWV) forecasting as a fully operational tool for optimizing astronomical infrared (IR) observations at Roque de los Muchachos Observatory (ORM). For the model validation we used GNSS-based (Global Navigation Satellite System) data from the PWV monitor located at the ORM. We have run WRF every 24 h for near two months, with a horizon of 48 hours (hourly forecasts), from 2016 January 11 to 2016 March 4. These runs represent 1296 hourly forecast points. The validation is carried out using different approaches: performance as a function of the forecast range, time horizon accuracy, performance as a function of the PWV value, and performance of the operational WRF time series with 24- and 48-hour horizons. Excellent agreement was found between the model forecasts and observations, with R =0.951 and R =0.904 for the 24- and 48-h forecast time series respectively. The 48-h forecast was further improved by correcting a time lag of 2 h found in the predictions. The final errors, taking into account all the uncertainties involved, are 1.75 mm for the 24-h forecasts and 1.99 mm for 48 h. We found linear trends in both the correlation and RMSE of the residuals (measurements - forecasts) as a function of the forecast range within the horizons analysed (up to 48 h). In summary, the WRF performance is excellent and accurate, thus allowing it to be implemented as an operational tool at the ORM.
Online short-term solar power forecasting
DEFF Research Database (Denmark)
Bacher, Peder; Madsen, Henrik; Nielsen, Henrik Aalborg
2009-01-01
This paper describes a new approach to online forecasting of power production from PV systems. The method is suited to online forecasting in many applications and in this paper it is used to predict hourly values of solar power for horizons of up to 36 hours. The data used is fifteen......-minute observations of solar power from 21 PV systems located on rooftops in a small village in Denmark. The suggested method is a two-stage method where first a statistical normalization of the solar power is obtained using a clear sky model. The clear sky model is found using statistical smoothing techniques....... Then forecasts of the normalized solar power are calculated using adaptive linear time series models. Both autoregressive (AR) and AR with exogenous input (ARX) models are evaluated, where the latter takes numerical weather predictions (NWPs) as input. The results indicate that for forecasts up to two hours...
Sea Level Forecasts Aggregated from Established Operational Systems
Directory of Open Access Journals (Sweden)
Andy Taylor
2017-08-01
Full Text Available A system for providing routine seven-day forecasts of sea level observable at tide gauge locations is described and evaluated. Forecast time series are aggregated from well-established operational systems of the Australian Bureau of Meteorology; although following some adjustments these systems are only quasi-complimentary. Target applications are routine coastal decision processes under non-extreme conditions. The configuration aims to be relatively robust to operational realities such as version upgrades, data gaps and metadata ambiguities. Forecast skill is evaluated against hourly tide gauge observations. Characteristics of the bias correction term are demonstrated to be primarily static in time, with time varying signals showing regional coherence. This simple approach to exploiting existing complex systems can offer valuable levels of skill at a range of Australian locations. The prospect of interpolation between observation sites and exploitation of lagged-ensemble uncertainty estimates could be meaningfully pursued. Skill characteristics define a benchmark against which new operational sea level forecasting systems can be measured. More generally, an aggregation approach may prove to be optimal for routine sea level forecast services given the physically inhomogeneous processes involved and ability to incorporate ongoing improvements and extensions of source systems.
Kolmogorov Space in Time Series Data
Kanjamapornkul, K.; Pinčák, R.
2016-01-01
We provide the proof that the space of time series data is a Kolmogorov space with $T_{0}$-separation axiom using the loop space of time series data. In our approach we define a cyclic coordinate of intrinsic time scale of time series data after empirical mode decomposition. A spinor field of time series data comes from the rotation of data around price and time axis by defining a new extradimension to time series data. We show that there exist hidden eight dimensions in Kolmogorov space for ...
Short-Term Wind Speed Forecasting for Power System Operations
Zhu, Xinxin
2012-04-01
The emphasis on renewable energy and concerns about the environment have led to large-scale wind energy penetration worldwide. However, there are also significant challenges associated with the use of wind energy due to the intermittent and unstable nature of wind. High-quality short-term wind speed forecasting is critical to reliable and secure power system operations. This article begins with an overview of the current status of worldwide wind power developments and future trends. It then reviews some statistical short-term wind speed forecasting models, including traditional time series approaches and more advanced space-time statistical models. It also discusses the evaluation of forecast accuracy, in particular, the need for realistic loss functions. New challenges in wind speed forecasting regarding ramp events and offshore wind farms are also presented. © 2012 The Authors. International Statistical Review © 2012 International Statistical Institute.
Short-term heat load forecasting for single family houses
DEFF Research Database (Denmark)
Bacher, Peder; Madsen, Henrik; Nielsen, Henrik Aalborg
2013-01-01
This paper presents a method for forecasting the load for space heating in a single-family house. The forecasting model is built using data from sixteen houses located in Sønderborg, Denmark, combined with local climate measurements and weather forecasts. Every hour the hourly heat load for each...... house the following two days is forecasted. The forecast models are adaptive linear time-series models and the climate inputs used are: ambient temperature, global radiation and wind speed. A computationally efficient recursive least squares scheme is used. The models are optimized to fit the individual...... noise and that practically all correlation to the climate variables are removed. Furthermore, the results show that the forecasting errors mainly are related to: unpredictable high frequency variations in the heat load signal (predominant only for some houses), shifts in resident behavior patterns...
Trottini, Mario; Vigo, Isabel; Belda, Santiago
2015-01-01
Given a time series, running trends analysis (RTA) involves evaluating least squares trends over overlapping time windows of L consecutive time points, with overlap by all but one observation. This produces a new series called the “running trends series,” which is used as summary statistics of the original series for further analysis. In recent years, RTA has been widely used in climate applied research as summary statistics for time series and time series association. There is no doubt that ...
Directory of Open Access Journals (Sweden)
C. Serio
1997-06-01
Full Text Available The time dynamics of geoelectrical precursory time series has been investigated and a method to discriminate chaotic behaviour in geoelectrical precursory time series is proposed. It allows us to detect low-dimensional chaos when the only information about the time series comes from the time series themselves. The short-term predictability of these time series is evaluated using two possible forecasting approaches: global autoregressive approximation and local autoregressive approximation. The first views the data as a realization of a linear stochastic process, whereas the second considers the data points as a realization of a deterministic process, supposedly non-linear. The comparison of the predictive skill of the two techniques is a test to discriminate between low-dimensional chaos and random dynamics. The analyzed time series are geoelectrical measurements recorded by an automatic station located in Tito (Southern Italy in one of the most seismic areas of the Mediterranean region. Our findings are that the global (linear approach is superior to the local one and the physical system governing the phenomena of electrical nature is characterized by a large number of degrees of freedom. Power spectra of the filtered time series follow a P(f = F-a scaling law: they exhibit the typical behaviour of a broad class of fractal stochastic processes and they are a signature of the self-organized systems.
In-season retail sales forecasting using survival models
African Journals Online (AJOL)
Retail sales forecasting, survival analysis, time series analysis, Holt's smoothing .... where fx(t) is the probability density function of the future lifetime, Tx, of a .... Adjustments were made to the shape of the smoothed mortality rates in light of new.
Multiple Indicator Stationary Time Series Models.
Sivo, Stephen A.
2001-01-01
Discusses the propriety and practical advantages of specifying multivariate time series models in the context of structural equation modeling for time series and longitudinal panel data. For time series data, the multiple indicator model specification improves on classical time series analysis. For panel data, the multiple indicator model…
Use of flood propagation models in real time hydrologic forecast: experiences at Segura River
International Nuclear Information System (INIS)
Valverde, Angel Luis Aldana; Beato, Ana Martinez Perez
2004-01-01
In this paper a case study related to flood propagation forecast in the Segura River in Spain is presented along with the application that was developed for that purpose. Simulation and forecast models ease the work carry out by the watershed organism personnel and may be essential to understand the complexity of some of the propagation phenomena that take place at specific locations such as the study area, a man-made channel at the downstream end of the Segura River (from Contraparada to Guardamar), including the tributaries along the stream. Three different models were used in the previous studies: a steady state numerical model (Hec-Ras), a physical model and two unsteady state numerical models (ISIS and HMS). Also, historical time series were analyzed and some topography works were carried out along the stream. PROC Segura model was conceived for real time flood propagation forecast in the mentioned area using the data collected by the SAIH. A simplified model was developed based on the following methods: Muskingum, Muskingum-Cunge and Modified Puls. To overcome some of these models limitations, such as the one to one discharge-water surface relationships and the impossibility of reproducing downstream backwater, doubled input rating curves were used to estimate the discharge at some of the gauging stations located at the tributaries, i.e. Merancho and Rambia del Derramador, which may be affected by the water level in the Segura River. The advantages of using these simplified models versus a dynamic wave model were studied and reported as well. In general, it can be stated that when several solutions are provided to solve the same problem, the simplest solution is usually the best one.(Author)
Forecasting Natural Rubber Price In Malaysia Using Arima
Zahari, Fatin Z.; Khalid, Kamil; Roslan, Rozaini; Sufahani, Suliadi; Mohamad, Mahathir; Saifullah Rusiman, Mohd; Ali, Maselan
2018-04-01
This paper contains introduction, materials and methods, results and discussions, conclusions and references. Based on the title mentioned, high volatility of the price of natural rubber nowadays will give the significant risk to the producers, traders, consumers, and others parties involved in the production of natural rubber. To help them in making decisions, forecasting is needed to predict the price of natural rubber. The main objective of the research is to forecast the upcoming price of natural rubber by using the reliable statistical method. The data are gathered from Malaysia Rubber Board which the data are from January 2000 until December 2015. In this research, average monthly price of Standard Malaysia Rubber 20 (SMR20) will be forecast by using Box-Jenkins approach. Time series plot is used to determine the pattern of the data. The data have trend pattern which indicates the data is non-stationary data and the data need to be transformed. By using the Box-Jenkins method, the best fit model for the time series data is ARIMA (1, 1, 0) which this model satisfy all the criteria needed. Hence, ARIMA (1, 1, 0) is the best fitted model and the model will be used to forecast the average monthly price of Standard Malaysia Rubber 20 (SMR20) for twelve months ahead.
Interevent times in a new alarm-based earthquake forecasting model
Talbi, Abdelhak; Nanjo, Kazuyoshi; Zhuang, Jiancang; Satake, Kenji; Hamdache, Mohamed
2013-09-01
This study introduces a new earthquake forecasting model that uses the moment ratio (MR) of the first to second order moments of earthquake interevent times as a precursory alarm index to forecast large earthquake events. This MR model is based on the idea that the MR is associated with anomalous long-term changes in background seismicity prior to large earthquake events. In a given region, the MR statistic is defined as the inverse of the index of dispersion or Fano factor, with MR values (or scores) providing a biased estimate of the relative regional frequency of background events, here termed the background fraction. To test the forecasting performance of this proposed MR model, a composite Japan-wide earthquake catalogue for the years between 679 and 2012 was compiled using the Japan Meteorological Agency catalogue for the period between 1923 and 2012, and the Utsu historical seismicity records between 679 and 1922. MR values were estimated by sampling interevent times from events with magnitude M ≥ 6 using an earthquake random sampling (ERS) algorithm developed during previous research. Three retrospective tests of M ≥ 7 target earthquakes were undertaken to evaluate the long-, intermediate- and short-term performance of MR forecasting, using mainly Molchan diagrams and optimal spatial maps obtained by minimizing forecasting error defined by miss and alarm rate addition. This testing indicates that the MR forecasting technique performs well at long-, intermediate- and short-term. The MR maps produced during long-term testing indicate significant alarm levels before 15 of the 18 shallow earthquakes within the testing region during the past two decades, with an alarm region covering about 20 per cent (alarm rate) of the testing region. The number of shallow events missed by forecasting was reduced by about 60 per cent after using the MR method instead of the relative intensity (RI) forecasting method. At short term, our model succeeded in forecasting the
The development rainfall forecasting using kalman filter
Zulfi, Mohammad; Hasan, Moh.; Dwidja Purnomo, Kosala
2018-04-01
Rainfall forecasting is very interesting for agricultural planing. Rainfall information is useful to make decisions about the plan planting certain commodities. In this studies, the rainfall forecasting by ARIMA and Kalman Filter method. Kalman Filter method is used to declare a time series model of which is shown in the form of linear state space to determine the future forecast. This method used a recursive solution to minimize error. The rainfall data in this research clustered by K-means clustering. Implementation of Kalman Filter method is for modelling and forecasting rainfall in each cluster. We used ARIMA (p,d,q) to construct a state space for KalmanFilter model. So, we have four group of the data and one model in each group. In conclusions, Kalman Filter method is better than ARIMA model for rainfall forecasting in each group. It can be showed from error of Kalman Filter method that smaller than error of ARIMA model.
Real-time forecasts of flood hazard and impact: some UK experiences
Directory of Open Access Journals (Sweden)
Cole Steven J.
2016-01-01
Full Text Available Major UK floods over the last decade have motivated significant technological and scientific advances in operational flood forecasting and warning. New joint forecasting centres between the national hydrological and meteorological operating agencies have been formed that issue a daily, national Flood Guidance Statement (FGS to the emergency response community. The FGS is based on a Flood Risk Matrix approach that is a function of potential impact severity and likelihood. It has driven an increased demand for robust, accurate and timely forecast and alert information on fluvial and surface water flooding along with impact assessments. The Grid-to-Grid (G2G distributed hydrological model has been employed across Britain at a 1km resolution to support the FGS. Novel methods for linking dynamic gridded estimates of river flow and surface runoff with more detailed offline flood risk maps have been developed to obtain real-time probabilistic forecasts of potential impacts, leading to operational trials. Examples of the national-scale G2G application are provided along with case studies of forecast flood impact from (i an operational Surface Water Flooding (SWF trial during the Glasgow 2014 Commonwealth Games, (ii SWF developments under the Natural Hazards Partnership over England & Wales, and (iii fluvial applications in Scotland.
Real-Time Forecasting of Echo-Centroid Motion.
1979-01-01
is apparent that after five observations are obtained, the forecast error drops considerably. The normal lifetime of an echo (25 to 30 min) is...10kmI I ! Fig. 11. Track of 5 April 1978 mesocyclone (M) and two TVS’s (1) and (2). Times are CST. Pumpkin Center tornado is hatched and Marlow tornado is
Earthquake and failure forecasting in real-time: A Forecasting Model Testing Centre
Filgueira, Rosa; Atkinson, Malcolm; Bell, Andrew; Main, Ian; Boon, Steven; Meredith, Philip
2013-04-01
Across Europe there are a large number of rock deformation laboratories, each of which runs many experiments. Similarly there are a large number of theoretical rock physicists who develop constitutive and computational models both for rock deformation and changes in geophysical properties. Here we consider how to open up opportunities for sharing experimental data in a way that is integrated with multiple hypothesis testing. We present a prototype for a new forecasting model testing centre based on e-infrastructures for capturing and sharing data and models to accelerate the Rock Physicist (RP) research. This proposal is triggered by our work on data assimilation in the NERC EFFORT (Earthquake and Failure Forecasting in Real Time) project, using data provided by the NERC CREEP 2 experimental project as a test case. EFFORT is a multi-disciplinary collaboration between Geoscientists, Rock Physicists and Computer Scientist. Brittle failure of the crust is likely to play a key role in controlling the timing of a range of geophysical hazards, such as volcanic eruptions, yet the predictability of brittle failure is unknown. Our aim is to provide a facility for developing and testing models to forecast brittle failure in experimental and natural data. Model testing is performed in real-time, verifiably prospective mode, in order to avoid selection biases that are possible in retrospective analyses. The project will ultimately quantify the predictability of brittle failure, and how this predictability scales from simple, controlled laboratory conditions to the complex, uncontrolled real world. Experimental data are collected from controlled laboratory experiments which includes data from the UCL Laboratory and from Creep2 project which will undertake experiments in a deep-sea laboratory. We illustrate the properties of the prototype testing centre by streaming and analysing realistically noisy synthetic data, as an aid to generating and improving testing methodologies in
DEFF Research Database (Denmark)
Moskowitz, Tobias J.; Ooi, Yao Hua; Heje Pedersen, Lasse
2012-01-01
We document significant “time series momentum” in equity index, currency, commodity, and bond futures for each of the 58 liquid instruments we consider. We find persistence in returns for one to 12 months that partially reverses over longer horizons, consistent with sentiment theories of initial...... under-reaction and delayed over-reaction. A diversified portfolio of time series momentum strategies across all asset classes delivers substantial abnormal returns with little exposure to standard asset pricing factors and performs best during extreme markets. Examining the trading activities...
A Pro-active Real-time Forecasting and Decision Support System for Daily Management of Marine Works
Bollen, Mark; Leyssen, Gert; Smets, Steven; De Wachter, Tom
2016-04-01
this toolbox in real-time situations and facilitate forecasting of impacts of planned dredge works, the following operational online functionalities are implemented: • Automated fetch and preparation of the input data, including 7 day forecast wind and wave fields and real-time measurements, and user defined the turbidity inputs based on scheduled marine works. • Generate automated forecasts and running user configurable scenarios at the same time in parallel. • Export and convert the model results, time series and maps, into a standardized format (netcdf). • Automatic analysis and processing of model results, including the calculation of indicator turbidity values and the exceedance analysis of threshold levels at the different sensitive areas. Data assimilation with the real time on site turbidity measurements is implemented in this threshold analysis. • Pre-programmed generation of animated sediment plumes, specific charts and pdf reports to allow a rapid interpretation of the model results by the operators and facilitating decision making in the operational planning. The performed marine works, resulting from the marine work schedule proposed by the forecasting system, are evaluated by a threshold analysis on the validated turbidity measurements on the sensitive sites. This machine learning loop allows a check of the system in order to evaluate forecast and model uncertainties.
Kwasniok, Frank
2013-11-01
A time series analysis method for predicting the probability density of a dynamical system is proposed. A nonstationary parametric model of the probability density is estimated from data within a maximum likelihood framework and then extrapolated to forecast the future probability density and explore the system for critical transitions or tipping points. A full systematic account of parameter uncertainty is taken. The technique is generic, independent of the underlying dynamics of the system. The method is verified on simulated data and then applied to prediction of Arctic sea-ice extent.
Hybrid ellipsoidal fuzzy systems in forecasting regional electricity loads
Energy Technology Data Exchange (ETDEWEB)
Pai, Ping-Feng [Department of Information Management, National Chi Nan University, 1 University Road, Puli, Nantou 545, Taiwan (China)
2006-09-15
Because of the privatization of electricity in many countries, load forecasting has become one of the most crucial issues in the planning and operations of electric utilities. In addition, accurate regional load forecasting can provide the transmission and distribution operators with more information. The hybrid ellipsoidal fuzzy system was originally designed to solve control and pattern recognition problems. The main objective of this investigation is to develop a hybrid ellipsoidal fuzzy system for time series forecasting (HEFST) and apply the proposed model to forecast regional electricity loads in Taiwan. Additionally, a scaled conjugate gradient learning method is employed in the supervised learning phase of the HEFST model. Subsequently, numerical data taken from the existing literature is used to demonstrate the forecasting performance of the HEFST model. Simulation results reveal that the proposed model has better forecasting performance than the artificial neural network model and the regression model. Thus, the HEFST model is a valid and promising alternative for forecasting regional electricity loads. (author)
The Betting Odds Rating System: Using soccer forecasts to forecast soccer.
Wunderlich, Fabian; Memmert, Daniel
2018-01-01
Betting odds are frequently found to outperform mathematical models in sports related forecasting tasks, however the factors contributing to betting odds are not fully traceable and in contrast to rating-based forecasts no straightforward measure of team-specific quality is deducible from the betting odds. The present study investigates the approach of combining the methods of mathematical models and the information included in betting odds. A soccer forecasting model based on the well-known ELO rating system and taking advantage of betting odds as a source of information is presented. Data from almost 15.000 soccer matches (seasons 2007/2008 until 2016/2017) are used, including both domestic matches (English Premier League, German Bundesliga, Spanish Primera Division and Italian Serie A) and international matches (UEFA Champions League, UEFA Europe League). The novel betting odds based ELO model is shown to outperform classic ELO models, thus demonstrating that betting odds prior to a match contain more relevant information than the result of the match itself. It is shown how the novel model can help to gain valuable insights into the quality of soccer teams and its development over time, thus having a practical benefit in performance analysis. Moreover, it is argued that network based approaches might help in further improving rating and forecasting methods.
Papacharalampous, Georgia; Tyralis, Hristos; Koutsoyiannis, Demetris
2017-04-01
Machine learning (ML) is considered to be a promising approach to hydrological processes forecasting. We conduct a comparison between several stochastic and ML point estimation methods by performing large-scale computational experiments based on simulations. The purpose is to provide generalized results, while the respective comparisons in the literature are usually based on case studies. The stochastic methods used include simple methods, models from the frequently used families of Autoregressive Moving Average (ARMA), Autoregressive Fractionally Integrated Moving Average (ARFIMA) and Exponential Smoothing models. The ML methods used are Random Forests (RF), Support Vector Machines (SVM) and Neural Networks (NN). The comparison refers to the multi-step ahead forecasting properties of the methods. A total of 20 methods are used, among which 9 are the ML methods. 12 simulation experiments are performed, while each of them uses 2 000 simulated time series of 310 observations. The time series are simulated using stochastic processes from the families of ARMA and ARFIMA models. Each time series is split into a fitting (first 300 observations) and a testing set (last 10 observations). The comparative assessment of the methods is based on 18 metrics, that quantify the methods' performance according to several criteria related to the accurate forecasting of the testing set, the capturing of its variation and the correlation between the testing and forecasted values. The most important outcome of this study is that there is not a uniformly better or worse method. However, there are methods that are regularly better or worse than others with respect to specific metrics. It appears that, although a general ranking of the methods is not possible, their classification based on their similar or contrasting performance in the various metrics is possible to some extent. Another important conclusion is that more sophisticated methods do not necessarily provide better forecasts
Multiresolution wavelet-ANN model for significant wave height forecasting.
Digital Repository Service at National Institute of Oceanography (India)
Deka, P.C.; Mandal, S.; Prahlada, R.
Hybrid wavelet artificial neural network (WLNN) has been applied in the present study to forecast significant wave heights (Hs). Here Discrete Wavelet Transformation is used to preprocess the time series data (Hs) prior to Artificial Neural Network...
Statistical Uncertainty Estimation Using Random Forests and Its Application to Drought Forecast
Chen, Junfei; Li, Ming; Wang, Weiguang
2012-01-01
Drought is part of natural climate variability and ranks the first natural disaster in the world. Drought forecasting plays an important role in mitigating impacts on agriculture and water resources. In this study, a drought forecast model based on the random forest method is proposed to predict the time series of monthly standardized precipitation index (SPI). We demonstrate model application by four stations in the Haihe river basin, China. The random-forest- (RF-) based forecast model has ...
Demand Forecasting at Low Aggregation Levels using Factored Conditional Restricted Boltzmann Machine
DEFF Research Database (Denmark)
Mocanu, Elena; Nguyen, Phuong H.; Gibescu, Madeleine
2016-01-01
electric power consumption, local price and meteorological data collected from 1900 customers. The households are equipped with local generation and smart appliances capable of responding to realtime pricing signals. The results show that for the short-term (5 minute to 1 day ahead) prediction problems......The electrical demand forecasting problem can be regarded as a nonlinear time series prediction problem depending on many complex factors since it is required at various aggregation levels and at high temporal resolution. To solve this challenging problem, various time series and machine learning...... developed deep learning model for time series prediction, namely Factored Conditional Restricted Boltzmann Machine (FCRBM), and extend it for electrical demand forecasting. The assessment is made on the EcoGrid dataset, originating from the Bornholm island experiment in Denmark, consisting of aggregated...
Deterministic Echo State Networks Based Stock Price Forecasting
Directory of Open Access Journals (Sweden)
Jingpei Dan
2014-01-01
Full Text Available Echo state networks (ESNs, as efficient and powerful computational models for approximating nonlinear dynamical systems, have been successfully applied in financial time series forecasting. Reservoir constructions in standard ESNs rely on trials and errors in real applications due to a series of randomized model building stages. A novel form of ESN with deterministically constructed reservoir is competitive with standard ESN by minimal complexity and possibility of optimizations for ESN specifications. In this paper, forecasting performances of deterministic ESNs are investigated in stock price prediction applications. The experiment results on two benchmark datasets (Shanghai Composite Index and S&P500 demonstrate that deterministic ESNs outperform standard ESN in both accuracy and efficiency, which indicate the prospect of deterministic ESNs for financial prediction.
Combined time-varying forecast based on the proper scoring approach for wind power generation
DEFF Research Database (Denmark)
Chen, Xingying; Jiang, Yu; Yu, Kun
2017-01-01
Compared with traditional point forecasts, combined forecast have been proposed as an effective method to provide more accurate forecasts than individual model. However, the literature and research focus on wind-power combined forecasts are relatively limited. Here, based on forecasting error...... distribution, a proper scoring approach is applied to combine plausible models to form an overall time-varying model for the next day forecasts, rather than weights-based combination. To validate the effectiveness of the proposed method, real data of 3 years were used for testing. Simulation results...... demonstrate that the proposed method improves the accuracy of overall forecasts, even compared with a numerical weather prediction....
Online probabilistic learning with an ensemble of forecasts
Thorey, Jean; Mallet, Vivien; Chaussin, Christophe
2016-04-01
Our objective is to produce a calibrated weighted ensemble to forecast a univariate time series. In addition to a meteorological ensemble of forecasts, we rely on observations or analyses of the target variable. The celebrated Continuous Ranked Probability Score (CRPS) is used to evaluate the probabilistic forecasts. However applying the CRPS on weighted empirical distribution functions (deriving from the weighted ensemble) may introduce a bias because of which minimizing the CRPS does not produce the optimal weights. Thus we propose an unbiased version of the CRPS which relies on clusters of members and is strictly proper. We adapt online learning methods for the minimization of the CRPS. These methods generate the weights associated to the members in the forecasted empirical distribution function. The weights are updated before each forecast step using only past observations and forecasts. Our learning algorithms provide the theoretical guarantee that, in the long run, the CRPS of the weighted forecasts is at least as good as the CRPS of any weighted ensemble with weights constant in time. In particular, the performance of our forecast is better than that of any subset ensemble with uniform weights. A noteworthy advantage of our algorithm is that it does not require any assumption on the distributions of the observations and forecasts, both for the application and for the theoretical guarantee to hold. As application example on meteorological forecasts for photovoltaic production integration, we show that our algorithm generates a calibrated probabilistic forecast, with significant performance improvements on probabilistic diagnostic tools (the CRPS, the reliability diagram and the rank histogram).
Flood forecasting and uncertainty of precipitation forecasts
International Nuclear Information System (INIS)
Kobold, Mira; Suselj, Kay
2004-01-01
The timely and accurate flood forecasting is essential for the reliable flood warning. The effectiveness of flood warning is dependent on the forecast accuracy of certain physical parameters, such as the peak magnitude of the flood, its timing, location and duration. The conceptual rainfall - runoff models enable the estimation of these parameters and lead to useful operational forecasts. The accurate rainfall is the most important input into hydrological models. The input for the rainfall can be real time rain-gauges data, or weather radar data, or meteorological forecasted precipitation. The torrential nature of streams and fast runoff are characteristic for the most of the Slovenian rivers. Extensive damage is caused almost every year- by rainstorms affecting different regions of Slovenia' The lag time between rainfall and runoff is very short for Slovenian territory and on-line data are used only for now casting. Forecasted precipitations are necessary for hydrological forecast for some days ahead. ECMWF (European Centre for Medium-Range Weather Forecasts) gives general forecast for several days ahead while more detailed precipitation data with limited area ALADIN/Sl model are available for two days ahead. There is a certain degree of uncertainty using such precipitation forecasts based on meteorological models. The variability of precipitation is very high in Slovenia and the uncertainty of ECMWF predicted precipitation is very large for Slovenian territory. ECMWF model can predict precipitation events correctly, but underestimates amount of precipitation in general The average underestimation is about 60% for Slovenian region. The predictions of limited area ALADIN/Si model up to; 48 hours ahead show greater applicability in hydrological forecasting. The hydrological models are sensitive to precipitation input. The deviation of runoff is much bigger than the rainfall deviation. Runoff to rainfall error fraction is about 1.6. If spatial and time distribution
Wu, Qi
2010-03-01
Demand forecasts play a crucial role in supply chain management. The future demand for a certain product is the basis for the respective replenishment systems. Aiming at demand series with small samples, seasonal character, nonlinearity, randomicity and fuzziness, the existing support vector kernel does not approach the random curve of the sales time series in the space (quadratic continuous integral space). In this paper, we present a hybrid intelligent system combining the wavelet kernel support vector machine and particle swarm optimization for demand forecasting. The results of application in car sale series forecasting show that the forecasting approach based on the hybrid PSOWv-SVM model is effective and feasible, the comparison between the method proposed in this paper and other ones is also given, which proves that this method is, for the discussed example, better than hybrid PSOv-SVM and other traditional methods.
Omar, Hani; Hoang, Van Hai; Liu, Duen-Ren
2016-01-01
Enhancing sales and operations planning through forecasting analysis and business intelligence is demanded in many industries and enterprises. Publishing industries usually pick attractive titles and headlines for their stories to increase sales, since popular article titles and headlines can attract readers to buy magazines. In this paper, information retrieval techniques are adopted to extract words from article titles. The popularity measures of article titles are then analyzed by using the search indexes obtained from Google search engine. Backpropagation Neural Networks (BPNNs) have successfully been used to develop prediction models for sales forecasting. In this study, we propose a novel hybrid neural network model for sales forecasting based on the prediction result of time series forecasting and the popularity of article titles. The proposed model uses the historical sales data, popularity of article titles, and the prediction result of a time series, Autoregressive Integrated Moving Average (ARIMA) forecasting method to learn a BPNN-based forecasting model. Our proposed forecasting model is experimentally evaluated by comparing with conventional sales prediction techniques. The experimental result shows that our proposed forecasting method outperforms conventional techniques which do not consider the popularity of title words.
Omar, Hani; Hoang, Van Hai; Liu, Duen-Ren
2016-01-01
Enhancing sales and operations planning through forecasting analysis and business intelligence is demanded in many industries and enterprises. Publishing industries usually pick attractive titles and headlines for their stories to increase sales, since popular article titles and headlines can attract readers to buy magazines. In this paper, information retrieval techniques are adopted to extract words from article titles. The popularity measures of article titles are then analyzed by using the search indexes obtained from Google search engine. Backpropagation Neural Networks (BPNNs) have successfully been used to develop prediction models for sales forecasting. In this study, we propose a novel hybrid neural network model for sales forecasting based on the prediction result of time series forecasting and the popularity of article titles. The proposed model uses the historical sales data, popularity of article titles, and the prediction result of a time series, Autoregressive Integrated Moving Average (ARIMA) forecasting method to learn a BPNN-based forecasting model. Our proposed forecasting model is experimentally evaluated by comparing with conventional sales prediction techniques. The experimental result shows that our proposed forecasting method outperforms conventional techniques which do not consider the popularity of title words. PMID:27313605
Stochastic models for time series
Doukhan, Paul
2018-01-01
This book presents essential tools for modelling non-linear time series. The first part of the book describes the main standard tools of probability and statistics that directly apply to the time series context to obtain a wide range of modelling possibilities. Functional estimation and bootstrap are discussed, and stationarity is reviewed. The second part describes a number of tools from Gaussian chaos and proposes a tour of linear time series models. It goes on to address nonlinearity from polynomial or chaotic models for which explicit expansions are available, then turns to Markov and non-Markov linear models and discusses Bernoulli shifts time series models. Finally, the volume focuses on the limit theory, starting with the ergodic theorem, which is seen as the first step for statistics of time series. It defines the distributional range to obtain generic tools for limit theory under long or short-range dependences (LRD/SRD) and explains examples of LRD behaviours. More general techniques (central limit ...
Rodgers, Joseph Lee; Beasley, William Howard; Schuelke, Matthew
2014-01-01
Many data structures, particularly time series data, are naturally seasonal, cyclical, or otherwise circular. Past graphical methods for time series have focused on linear plots. In this article, we move graphical analysis onto the circle. We focus on 2 particular methods, one old and one new. Rose diagrams are circular histograms and can be produced in several different forms using the RRose software system. In addition, we propose, develop, illustrate, and provide software support for a new circular graphical method, called Wrap-Around Time Series Plots (WATS Plots), which is a graphical method useful to support time series analyses in general but in particular in relation to interrupted time series designs. We illustrate the use of WATS Plots with an interrupted time series design evaluating the effect of the Oklahoma City bombing on birthrates in Oklahoma County during the 10 years surrounding the bombing of the Murrah Building in Oklahoma City. We compare WATS Plots with linear time series representations and overlay them with smoothing and error bands. Each method is shown to have advantages in relation to the other; in our example, the WATS Plots more clearly show the existence and effect size of the fertility differential.
Road Short-Term Travel Time Prediction Method Based on Flow Spatial Distribution and the Relations
Directory of Open Access Journals (Sweden)
Mingjun Deng
2016-01-01
Full Text Available There are many short-term road travel time forecasting studies based on time series, but indeed, road travel time not only relies on the historical travel time series, but also depends on the road and its adjacent sections history flow. However, few studies have considered that. This paper is based on the correlation of flow spatial distribution and the road travel time series, applying nearest neighbor and nonparametric regression method to build a forecasting model. In aspect of spatial nearest neighbor search, three different space distances are defined. In addition, two forecasting functions are introduced: one combines the forecasting value by mean weight and the other uses the reciprocal of nearest neighbors distance as combined weight. Three different distances are applied in nearest neighbor search, which apply to the two forecasting functions. For travel time series, the nearest neighbor and nonparametric regression are applied too. Then minimizing forecast error variance is utilized as an objective to establish the combination model. The empirical results show that the combination model can improve the forecast performance obviously. Besides, the experimental results of the evaluation for the computational complexity show that the proposed method can satisfy the real-time requirement.
Directory of Open Access Journals (Sweden)
Gonghao Duan
2018-01-01
Full Text Available The loss of lake area significantly influences the climate change in a region, and this loss represents a serious and unavoidable challenge to maintaining ecological sustainability under the circumstances of lakes that are being filled. Therefore, mapping and forecasting changes in the lake is critical for protecting the environment and mitigating ecological problems in the urban district. We created an accessible map displaying area changes for 82 lakes in the Wuhan city using remote sensing data in conjunction with visual interpretation by combining field data with Landsat 2/5/7/8 Thematic Mapper (TM time-series images for the period 1987–2013. In addition, we applied a quadratic exponential smoothing model to forecast lake area changes in Wuhan city. The map provides, for the first time, estimates of lake development in Wuhan using data required for local-scale studies. The model predicted a lake area reduction of 18.494 km2 in 2015. The average error reached 0.23 with a correlation coefficient of 0.98, indicating that the model is reliable. The paper provided a numerical analysis and forecasting method to provide a better understanding of lake area changes. The modeling and mapping results can help assess aquatic habitat suitability and property planning for Wuhan lakes.
Kvetan, Vladimir, Ed.
2014-01-01
Reliable and consistent time series are essential to any kind of economic forecasting. Skills forecasting needs to combine data from national accounts and labour force surveys, with the pan-European dimension of Cedefop's skills supply and demand forecasts, relying on different international classification standards. Sectoral classification (NACE)…
Incorporating geostrophic wind information for improved space–time short-term wind speed forecasting
Zhu, Xinxin; Bowman, Kenneth P.; Genton, Marc G.
2014-01-01
pressure, temperature, and other meteorological variables, no improvement in forecasting accuracy was found by incorporating air pressure and temperature directly into an advanced space-time statistical forecasting model, the trigonometric direction diurnal
Uranium price forecasting methods
International Nuclear Information System (INIS)
Fuller, D.M.
1994-01-01
This article reviews a number of forecasting methods that have been applied to uranium prices and compares their relative strengths and weaknesses. The methods reviewed are: (1) judgemental methods, (2) technical analysis, (3) time-series methods, (4) fundamental analysis, and (5) econometric methods. Historically, none of these methods has performed very well, but a well-thought-out model is still useful as a basis from which to adjust to new circumstances and try again
西埜, 晴久
2004-01-01
The paper investigates an application of long-memory processes to economic time series. We show properties of long-memory processes, which are motivated to model a long-memory phenomenon in economic time series. An FARIMA model is described as an example of long-memory model in statistical terms. The paper explains basic limit theorems and estimation methods for long-memory processes in order to apply long-memory models to economic time series.
Results on SSH neural network forecasting in the Mediterranean Sea
Rixen, Michel; Beckers, Jean-Marie; Alvarez, Alberto; Tintore, Joaquim
2002-01-01
Nowadays, satellites are the only monitoring systems that cover almost continuously all possible ocean areas and are now an essential part of operational oceanography. A novel approach based on artificial intelligence (AI) concepts, exploits pasts time series of satellite images to infer near future ocean conditions at the surface by neural networks and genetic algorithms. The size of the AI problem is drastically reduced by splitting the spatio-temporal variability contained in the remote sensing data by using empirical orthogonal function (EOF) decomposition. The problem of forecasting the dynamics of a 2D surface field can thus be reduced by selecting the most relevant empirical modes, and non-linear time series predictors are then applied on the amplitudes only. In the present case study, we use altimetric maps of the Mediterranean Sea, combining TOPEX-POSEIDON and ERS-1/2 data for the period 1992 to 1997. The learning procedure is applied to each mode individually. The final forecast is then reconstructed form the EOFs and the forecasted amplitudes and compared to the real observed field for validation of the method.
Visibility Graph Based Time Series Analysis.
Stephen, Mutua; Gu, Changgui; Yang, Huijie
2015-01-01
Network based time series analysis has made considerable achievements in the recent years. By mapping mono/multivariate time series into networks, one can investigate both it's microscopic and macroscopic behaviors. However, most proposed approaches lead to the construction of static networks consequently providing limited information on evolutionary behaviors. In the present paper we propose a method called visibility graph based time series analysis, in which series segments are mapped to visibility graphs as being descriptions of the corresponding states and the successively occurring states are linked. This procedure converts a time series to a temporal network and at the same time a network of networks. Findings from empirical records for stock markets in USA (S&P500 and Nasdaq) and artificial series generated by means of fractional Gaussian motions show that the method can provide us rich information benefiting short-term and long-term predictions. Theoretically, we propose a method to investigate time series from the viewpoint of network of networks.
Visibility Graph Based Time Series Analysis.
Directory of Open Access Journals (Sweden)
Mutua Stephen
Full Text Available Network based time series analysis has made considerable achievements in the recent years. By mapping mono/multivariate time series into networks, one can investigate both it's microscopic and macroscopic behaviors. However, most proposed approaches lead to the construction of static networks consequently providing limited information on evolutionary behaviors. In the present paper we propose a method called visibility graph based time series analysis, in which series segments are mapped to visibility graphs as being descriptions of the corresponding states and the successively occurring states are linked. This procedure converts a time series to a temporal network and at the same time a network of networks. Findings from empirical records for stock markets in USA (S&P500 and Nasdaq and artificial series generated by means of fractional Gaussian motions show that the method can provide us rich information benefiting short-term and long-term predictions. Theoretically, we propose a method to investigate time series from the viewpoint of network of networks.
Forecasting biodiversity in breeding birds using best practices
Taylor, Shawn D.; White, Ethan P.
2018-01-01
Biodiversity forecasts are important for conservation, management, and evaluating how well current models characterize natural systems. While the number of forecasts for biodiversity is increasing, there is little information available on how well these forecasts work. Most biodiversity forecasts are not evaluated to determine how well they predict future diversity, fail to account for uncertainty, and do not use time-series data that captures the actual dynamics being studied. We addressed these limitations by using best practices to explore our ability to forecast the species richness of breeding birds in North America. We used hindcasting to evaluate six different modeling approaches for predicting richness. Hindcasts for each method were evaluated annually for a decade at 1,237 sites distributed throughout the continental United States. All models explained more than 50% of the variance in richness, but none of them consistently outperformed a baseline model that predicted constant richness at each site. The best practices implemented in this study directly influenced the forecasts and evaluations. Stacked species distribution models and “naive” forecasts produced poor estimates of uncertainty and accounting for this resulted in these models dropping in the relative performance compared to other models. Accounting for observer effects improved model performance overall, but also changed the rank ordering of models because it did not improve the accuracy of the “naive” model. Considering the forecast horizon revealed that the prediction accuracy decreased across all models as the time horizon of the forecast increased. To facilitate the rapid improvement of biodiversity forecasts, we emphasize the value of specific best practices in making forecasts and evaluating forecasting methods. PMID:29441230
Forecasting biodiversity in breeding birds using best practices
Directory of Open Access Journals (Sweden)
David J. Harris
2018-02-01
Full Text Available Biodiversity forecasts are important for conservation, management, and evaluating how well current models characterize natural systems. While the number of forecasts for biodiversity is increasing, there is little information available on how well these forecasts work. Most biodiversity forecasts are not evaluated to determine how well they predict future diversity, fail to account for uncertainty, and do not use time-series data that captures the actual dynamics being studied. We addressed these limitations by using best practices to explore our ability to forecast the species richness of breeding birds in North America. We used hindcasting to evaluate six different modeling approaches for predicting richness. Hindcasts for each method were evaluated annually for a decade at 1,237 sites distributed throughout the continental United States. All models explained more than 50% of the variance in richness, but none of them consistently outperformed a baseline model that predicted constant richness at each site. The best practices implemented in this study directly influenced the forecasts and evaluations. Stacked species distribution models and “naive” forecasts produced poor estimates of uncertainty and accounting for this resulted in these models dropping in the relative performance compared to other models. Accounting for observer effects improved model performance overall, but also changed the rank ordering of models because it did not improve the accuracy of the “naive” model. Considering the forecast horizon revealed that the prediction accuracy decreased across all models as the time horizon of the forecast increased. To facilitate the rapid improvement of biodiversity forecasts, we emphasize the value of specific best practices in making forecasts and evaluating forecasting methods.
On the importance of the long-term seasonal component in day-ahead electricity price forecasting
International Nuclear Information System (INIS)
Nowotarski, Jakub; Weron, Rafał
2016-01-01
In day-ahead electricity price forecasting (EPF) the daily and weekly seasonalities are always taken into account, but the long-term seasonal component (LTSC) is believed to add unnecessary complexity to the already parameter-rich models and is generally ignored. Conducting an extensive empirical study involving state-of-the-art time series models we show that (i) decomposing a series of electricity prices into a LTSC and a stochastic component, (ii) modeling them independently and (iii) combining their forecasts can bring – contrary to a common belief – an accuracy gain compared to an approach in which a given time series model is calibrated to the prices themselves. - Highlights: • A new class of Seasonal Component AutoRegressive (SCAR) models is introduced. • Electricity prices are decomposed into a trend-seasonal and a stochastic component. • Both components are modeled independently, their forecasts are combined. • Significant accuracy gains can be achieved compared to commonly used approaches.
An Experimental Real-Time Ocean Nowcast/Forecast System for Intra America Seas
Ko, D. S.; Preller, R. H.; Martin, P. J.
2003-04-01
An experimental real-time Ocean Nowcast/Forecast System has been developed for the Intra America Seas (IASNFS). The area of coverage includes the Caribbean Sea, the Gulf of Mexico and the Straits of Florida. The system produces nowcast and up to 72 hours forecast the sea level variation, 3D ocean current, temperature and salinity fields. IASNFS consists an 1/24 degree (~5 km), 41-level sigma-z data-assimilating ocean model based on NCOM. For daily nowcast/forecast the model is restarted from previous nowcast. Once model is restarted it continuously assimilates the synthetic temperature/salinity profiles generated by a data analysis model called MODAS to produce nowcast. Real-time data come from satellite altimeter (GFO, TOPEX/Poseidon, ERS-2) sea surface height anomaly and AVHRR sea surface temperature. Three hourly surface heat fluxes, including solar radiation, wind stresses and sea level air pressure from NOGAPS/FNMOC are applied for surface forcing. Forecasts are produced with available NOGAPS forecasts. Once the nowcast/forecast are produced they are distributed through the Internet via the updated web pages. The open boundary conditions including sea surface elevation, transport, temperature, salinity and currents are provided by the NRL 1/8 degree Global NCOM which is operated daily. An one way coupling scheme is used to ingest those boundary conditions into the IAS model. There are 41 rivers with monthly discharges included in the IASNFS.
Spatial-temporal analysis of wind power forecast errors for West-Coast Norway
Energy Technology Data Exchange (ETDEWEB)
Revheim, Paal Preede; Beyer, Hans Georg [Agder Univ. (UiA), Grimstad (Norway). Dept. of Engineering Sciences
2012-07-01
In this paper the spatial-temporal structure of forecast errors for wind power in West-Coast Norway is analyzed. Starting on the qualitative analysis of the forecast error reduction, with respect to single site data, for the lumped conditions of groups of sites the spatial and temporal correlations of the wind power forecast errors within and between the same groups are studied in detail. Based on this, time-series regression models to be used to analytically describe the error reduction are set up. The models give an expected reduction in forecast error between 48.4% and 49%. (orig.)
Forecasting Aggregate Productivity using Information from Firm-level Data
Bartelsman, E.J.; Wolf, Z.
2014-01-01
In this paper, we explore whether information from firm-level data can improve forecasts of aggregate productivity growth. We generate firm-level productivity measures and aggregate them into time-series components that capture within-firm productivity and the productivity contribution of
Electricity demand load forecasting of the Hellenic power system using an ARMA model
Energy Technology Data Exchange (ETDEWEB)
Pappas, S.Sp. [ASPETE - School of Pedagogical and Technological Education Department of Electrical Engineering Educators N. Heraklion, 141 21 Athens (Greece); Ekonomou, L.; Chatzarakis, G.E.; Skafidas, P.D. [ASPETE-School of Pedagogical and Technological Education, Department of Electrical Engineering Educators, N. Heraklion, 141 21 Athens (Greece); Karampelas, P. [Hellenic American University, IT Department, 12 Kaplanon Str., 106 80 Athens (Greece); Karamousantas, D.C. [Technological Educational Institute of Kalamata, Antikalamos, 24 100 Kalamata (Greece); Katsikas, S.K. [University of Piraeus, Department of Technology Education and Digital Systems, 150 Androutsou St., 18 532 Piraeus (Greece)
2010-03-15
Effective modeling and forecasting requires the efficient use of the information contained in the available data so that essential data properties can be extracted and projected into the future. As far as electricity demand load forecasting is concerned time series analysis has the advantage of being statistically adaptive to data characteristics compared to econometric methods which quite often are subject to errors and uncertainties in model specification and knowledge of causal variables. This paper presents a new method for electricity demand load forecasting using the multi-model partitioning theory and compares its performance with three other well established time series analysis techniques namely Corrected Akaike Information Criterion (AICC), Akaike's Information Criterion (AIC) and Schwarz's Bayesian Information Criterion (BIC). The suitability of the proposed method is illustrated through an application to actual electricity demand load of the Hellenic power system, proving the reliability and the effectiveness of the method and making clear its usefulness in the studies that concern electricity consumption and electricity prices forecasts. (author)
Development and testing of an innovative short-term large wind ramp forecasting system
Energy Technology Data Exchange (ETDEWEB)
Zack, J.W. [AWS Truepower LLC, Troy, NY (United States)
2010-07-01
This PowerPoint presentation discussed a ramp forecasting tool designed for use in a region of Texas with a high wind-generating capacity. Large system-wide ramps frequently occur in the region, and curtailments are common due to transmission constraints. The average hourly load of the power system is 32,101 MW. Wind power capacity in the region is 9382 MW. However, actual production rarely exceeds 6500 MW due to the curtailments. The short-term ramp forecasting tool was designed to aid in grid management decisions for the 0-6 hour ahead period as well as to address issues related to wind farm time series data and the lack of situational awareness information. The tool provided rapid updates for grid point wind analysis with feature detection and tracking algorithms and a rapid update cycle model. The tool also featured a suite of web-based applications that included deterministic ramp even forecasts, power production time series forecasts, and situational awareness products that are updated every 15 minutes. A performance evaluation study of the tool was provided. tabs., figs.
Analysis and Synthesis of Load Forecasting Data for Renewable Integration Studies: Preprint
Energy Technology Data Exchange (ETDEWEB)
Steckler, N.; Florita, A.; Zhang, J.; Hodge, B. M.
2013-11-01
As renewable energy constitutes greater portions of the generation fleet, the importance of modeling uncertainty as part of integration studies also increases. In pursuit of optimal system operations, it is important to capture not only the definitive behavior of power plants, but also the risks associated with systemwide interactions. This research examines the dependence of load forecast errors on external predictor variables such as temperature, day type, and time of day. The analysis was utilized to create statistically relevant instances of sequential load forecasts with only a time series of historic, measured load available. The creation of such load forecasts relies on Bayesian techniques for informing and updating the model, thus providing a basis for networked and adaptive load forecast models in future operational applications.
Using adaptive network based fuzzy inference system to forecast regional electricity loads
International Nuclear Information System (INIS)
Ying, L.-C.; Pan, M.-C.
2008-01-01
Since accurate regional load forecasting is very important for improvement of the management performance of the electric industry, various regional load forecasting methods have been developed. The purpose of this study is to apply the adaptive network based fuzzy inference system (ANFIS) model to forecast the regional electricity loads in Taiwan and demonstrate the forecasting performance of this model. Based on the mean absolute percentage errors and statistical results, we can see that the ANFIS model has better forecasting performance than the regression model, artificial neural network (ANN) model, support vector machines with genetic algorithms (SVMG) model, recurrent support vector machines with genetic algorithms (RSVMG) model and hybrid ellipsoidal fuzzy systems for time series forecasting (HEFST) model. Thus, the ANFIS model is a promising alternative for forecasting regional electricity loads
Using adaptive network based fuzzy inference system to forecast regional electricity loads
Energy Technology Data Exchange (ETDEWEB)
Ying, Li-Chih [Department of Marketing Management, Central Taiwan University of Science and Technology, 11, Pu-tzu Lane, Peitun, Taichung City 406 (China); Pan, Mei-Chiu [Graduate Institute of Management Sciences, Nanhua University, 32, Chung Keng Li, Dalin, Chiayi 622 (China)
2008-02-15
Since accurate regional load forecasting is very important for improvement of the management performance of the electric industry, various regional load forecasting methods have been developed. The purpose of this study is to apply the adaptive network based fuzzy inference system (ANFIS) model to forecast the regional electricity loads in Taiwan and demonstrate the forecasting performance of this model. Based on the mean absolute percentage errors and statistical results, we can see that the ANFIS model has better forecasting performance than the regression model, artificial neural network (ANN) model, support vector machines with genetic algorithms (SVMG) model, recurrent support vector machines with genetic algorithms (RSVMG) model and hybrid ellipsoidal fuzzy systems for time series forecasting (HEFST) model. Thus, the ANFIS model is a promising alternative for forecasting regional electricity loads. (author)