Automatic lag selection in time series forecasting using multiple kernel learning

详细信息    查看全文

• 作者：Agus Widodo ; Indra Budi ; Belawati Widjaja
• 关键词：Time series ; Forecast combination ; Sliding window ; Lag selection ; Multiple kernel learning
• 刊名：International Journal of Machine Learning and Cybernetics
• 出版年：2016
• 出版时间：February 2016
• 年：2016
• 卷：7
• 期：1
• 页码：95-110
• 全文大小：968 KB
• 参考文献：1.Clements MP, Franses PH, Swanson NR (2004) Forecasting economic and financial time-series with non-linear models. Int J Forecast 20:169–183CrossRef
  2.González-Romera E, Jaramillo-Morán MÁ, Carmona-Fernández D (2006) Monthly electric energy demand forecasting based on trend extraction. IEEE Trans Power Syst 21:1946–1953CrossRef
  3.Makridakis SG, Wheelwright SC, Hyndman RJ (1998) Forecasting: methods and applications. Wiley, New York
  4.Cao L (2003) Support vector machines experts for time series forecasting. Neurocomputing 51:321–339CrossRef
  5.Zhang GP, Kline DM (2007) Quarterly time-series forecasting with neural networks. Neural Netw IEEE Trans 18:1800–1814CrossRef
  6.Kourentzes N, Crone SF (2008) Automatic modelling of neural networks for time series prediction—in search of a uniform methodology across varying time frequencies. In: Proceedings of the 2nd European Symposium Time Series Predict
  7.Crone SF, Kourentzes N (2009) Forecasting seasonal time series with multilayer perceptrons – an empirical evaluation of input vector specifications for deterministic seasonality. In: Proceedings of the 2009 international conference on data mining, DMIN 2009, Las Vegas. CSREA Press, pp 232–238
  8.Clemen T (1989) Combining forecasts: a review and annotated bibliography. Int J Forecast 5:559–583CrossRef
  9.Siwek K, Osowski S, Szupiluk R (2009) Ensemble neural network approach for accurate load forecasting in a power system. Int J Appl Math Comput Sci 19:303–315MATH CrossRef
  10.Huang C, Yang D, Chuang Y (2008) Application of wrapper approach and composite classifier to the stock trend prediction. Expert Syst Appl 34:2870–2878CrossRef
  11.Armstrong JS (1989) Combining forecasts. Int J Forecast 5:585–588CrossRef
  12.Poncela P, Rodríguez J, Sánchez-Mangas R, Senra E (2011) Forecast combination through dimension reduction techniques. Int J Forecast 27:224–237CrossRef
  13.Andrawis RR, Atiya AF, El-Shishiny H (2011) Combination of long term and short term forecasts, with application to tourism demand forecasting. Int J Forecast 27:870–886CrossRef
  14.Kourentzes N, Petropoulos F, Trapero JR (2014) Improving forecasting by estimating time series structural components across multiple frequencies. Int J Forecast 30:291–302CrossRef
  15.Cortes C (2011) Ensembles of Kernel Predictors. In: Proceedings of the 27th Conference Uncertainty Artificial Intelligence
  16.Lee W, Verzakov S, Duin RPW (2007) Kernel combination versus classifier combination. Multi Classification System Lecture Notes Computer Science, vol 4472, pp 22–31
  17.Kim H-C, Pang S, Je H-M, Kim D, Yang Bang S (2003) Constructing support vector machine ensemble. Pattern Recognit 36:2757–2767MATH CrossRef
  18.Rakotomamonjy A, Bach FR, Grandvalet Y, Canu S (2008) SimpleMKL. J Mach Learn Res 9:2491–2521MATH MathSciNet
  19.Bach FR, Lanckriet GRG, Jordan MI (2004) Multiple kernel learning, conic duality, and the SMO algorithm. Twenty-first Int Conf Mach Learn—ICML’04 6
  20.Gonen M, Alpaydin E (2011) Multiple kernel learning algorithms. J Mach Learn Res 12:2211–2268MathSciNet
  21.Yeh C, Huang C, Lee S (2011) Expert Systems with Applications A multiple-kernel support vector regression approach for stock market price forecasting q. Expert Syst Appl 38:2177–2186CrossRef
  22.Zhang X, Hu L, Wang Z (2010) Multiple kernel support vector regression for economic forecasting. In: 2010 international conference on management science and engineering, Melbourne. IEEE, pp 129–134
  23.Tschernig R, Yang L (2000) Nonparametric lag selection for time series. J Time Ser Anal 21:457–487MATH MathSciNet CrossRef
  24.Crone SF, Kourentzes N (2010) Feature selection for time series prediction—A combined filter and wrapper approach for neural networks. Neurocomputing 73:1923–1936CrossRef
  25.Simon G, Verleysen M (2006) Lag selection for regression models using high-dimensional mutual information. In: European symposium on artificial neural networks, Bruges, Belgium, pp 395–400
  26.Ribeiro GHT, Neto PSGDM, Cavalcanti GDC, Tsang IR (2011) Lag selection for time series forecasting using particle swarm optimization. The 2011 International Joint Conference, pp 2437–2444
  27.Davey N, Hunt SP, Frank RJ Time series prediction and neural networks. In: Proceedings of the 5th International Conference on Engineering Applications of Neural Networks (EANN’99), pp 3–8
  28.Leon F, Zaharia MH (2010) Stacked heterogeneous neural networks for time series forecasting. Math Probl Eng 2010:1–20CrossRef
  29.Yoshida S, Hatano K, Takimoto E (2011) Adaptive online prediction using weighted windows. IEICE Trans Inf Syst 94-D:1917–1923CrossRef
  30.Sharda R, Patil RB (1992) Connectionist approach to time series prediction: an empirical test. J Intell Manuf 3:317–323CrossRef
  31.Nelson M, Hill T, Remus T, O’Connor M (1999) Time series forecasting using NNs: should the data be deseasonalized first? J Forecast 8:359–367CrossRef
  32.Theodosiou M (2011) Forecasting monthly and quarterly time series using STL decomposition. Int J Forecast 27:1178–1195CrossRef
  33.Christodoulos C, Michalakelis C, Varoutas D (2010) Forecasting with limited data: combining ARIMA and diffusion models. Technol Forecast Soc Change 77:558–565CrossRef
  34.Dileep AD, Sekhar CC (2009) Representation and feature selection using multiple kernel learning. In: Proceedings International Joint Conference Neural Networks. Atlanta, Georgia, pp 717–722
  35.Foresti L, Tuia D, Timonin V, Kanevski M (2010) Time series input selection using multiple kernel learning: 28–30
  36.Crone SF, Hibon M, Nikolopoulos K (2011) Advances in forecasting with neural networks? Empirical evidence from the NN3 competition on time series prediction. Int J Forecast 27:635–660CrossRef
  37.Makridakis S, Hibon M (2000) The M3-Competition: results, conclusions and implications. Int J Forecast 16:451–476CrossRef
  38.Zien A (2008) Multiple Kernel Learning. In: Friedrich Miescher Lab. http://​raetschlab.​org/​lectures/​mkl-tutorial.​pdf . Accessed 14 Dec 2012
  39.Kloft M, Laskov P, Zien A (2010) Efficient and accurate Lp-norm multiple kernel learning. Neural Inf Proc Sys 22:997–1005
  40.Anderson DR (2004) Multimodel inference understanding AIC and BIC in model selection. Soc Methods Res 33:261–304CrossRef
  41.Hyndman RJ (2011) Statistical tests for variable selection. http://​robjhyndman.​com/​hyndsight/​tests2/​ . Accessed 5 Jan 2013
  42.Berrar DP, Sturgeon B, Bradbury I, Dubitzky W (2003) Microarray data integration and machine learning techniques for lung cancer survival prediction. In: Proceedings of the the International Conference of Critical Assessment of Microarray Data Analysis
  43.Napolitano F, Zhao Y, Moreira VM, Tagliaferri R, Kere J, Amato MD, Greco D (2013) Drug repositioning: a machine-learning approach through data integration. J Cheminform 5:1–9CrossRef
  44.Ozen A, Gönen M, Alpaydın E, Haliloğlu T (2009) Machine learning integration for predicting the effect of single. BMC Struct Biol 17:1–17
  45.Bucak SS, Member S, Jin R, Jain AK (2014) Multiple kernel learning for visual object recognition : a review. IEEE Trans Pattern Anal Mach Intell 36:1354–1369CrossRef
  46.Yu S, Falck T, Daemen A, Tranchevent L, Suykens JAK, Moor B De, Moreau Y (2010) L2-norm multiple kernel learning and its application to biomedical data fusion. BMC Bioinform 11:309–322CrossRef
  47.Hyndman RJ, Athanasopoulos G (2012) Forecasting: principles and practice. In: Online, Open Access Textb. https://​www.​otexts.​org/​fpp
  48.Ramasubramanian V (2007) Time series analysis. IASRI, Library Avenue, New Delhi
  49.Torres-reyna O (2012) Time series. In: Data and statistical services. Princeton University. http://​www.​princeton.​edu/​~otorres/​TS101.​pdf . Accessed 14 Dec 2013
  50.Pearson R (2011) Exploring data in engineering, the science and medicine. Oxford University Press, Oxford
  51.Nielsen ML (2012) Hampel filter. http://​www.​mathworks.​com/​matlabcentral/​fileexchange/​34795-outlier-detection-and-removal-hampel/​content/​hampel.​m . Accessed 27 Jan 2013
  52.Yeh Y, Lin T, Chung Y, Wang YF (2012) A novel multiple kernel learning framework for heterogeneous feature fusion and variable selection 14:563–574
  53.Wang X, Smith-miles K, Hyndman R (2009) Rule induction for forecasting method selection: meta-learning the characteristics of univariate time series. Neurocomputing 72:2581–2594CrossRef
  54.Hyndman RJ, Khandakar Y (2008) Automatic time series forecasting: the forecast package for R. J Stat Softw 27:1–22CrossRef
  55.Box GEP, Jenkins GM (1970) Time series analysis: forecasting and control. Wiley, San FranciscoMATH
  56.Hyndman RJ (2006) Another look at forecast-accuracy metrics for intermittent demand. Foresight 4:43–46
  57.Kourentzes N (2007) Exponential smoothing models. http://​nikolaos.​kourentzes.​com . Accessed 15 Nov 2012
  58.Hyndman RJ (2013) Forecasting without forecasters. In: Keynote lecture at the 2013 international symposium forecast, Seoul
  59.Wang X-Z, He Q, Chen D-G, Yeung D (2005) A genetic algorithm for solving the inverse problem of support vector machines. Neurocomputing 68:225–238CrossRef
  60.Wang X-Z, Lu S-X, Zhai J-H (2008) fast fuzzy multicategory svm based on support vector domain description. Int J Pattern Recognit Artif Intell 22:109–120CrossRef
  
• 作者单位：Agus Widodo (1)
  Indra Budi (2)
  Belawati Widjaja (2)
  
  1. Agency for the Assessment and Application of Technology, Jl. MH Thamrin 8, Jakarta, 10340, Indonesia
  2. Faculty of Computer Science, University of Indonesia, Depok, 16424, Indonesia
  
• 刊物类别：Engineering
• 刊物主题：Artificial Intelligence and Robotics
  Statistical Physics, Dynamical Systems and Complexity
  Computational Intelligence
  Control , Robotics, Mechatronics
  
• 出版者：Springer Berlin / Heidelberg
• ISSN：1868-808X

文摘

This paper reports the feasibility of employing the recent approach on kernel learning, namely the multiple kernel learning (MKL), for time series forecasting to automatically select the optimal lag length or size of sliding windows. MKL is an approach to choose suitable kernels from a given pool of kernels by exploring the combination of multiple kernels. In this paper, we extend the MKL capability to select the optimal size of sliding windows for time series domain by adopting the data integration approach which has been previously studied in the domain of image processing. In this study, each kernel represents the different lengths of time series lag. In addition, we also examine the feasibility of MKL for decomposed time series. We use the dataset from previous time series competitions as our benchmark. Our experimental results indicate that our approaches perform competitively compared to the previous methods using the same dataset. Furthermore, MKL may predict the detrended time series without explicitly computing the seasonality. The advantage of our method is in its ability in automatically selecting the optimal size of sliding windows and finding the pattern of time series. Keywords Time series Forecast combination Sliding window Lag selection Multiple kernel learning