Artificial neural networks in time series forecasting: a comparative analysis

Allende, Héctor   ; Moraga, Claudio   ; Salas, Rodrigo

Kybernetika, Tome 38 (2002) no. 6, pp. 685-707

Abstract (VO)

Artificial neural networks (ANN) have received a great deal of attention in many fields of engineering and science. Inspired by the study of brain architecture, ANN represent a class of non-linear models capable of learning from data. ANN have been applied in many areas where statistical methods are traditionally employed. They have been used in pattern recognition, classification, prediction and process control. The purpose of this paper is to discuss ANN and compare them to non-linear time series models. We begin exploring recent developments in time series forecasting with particular emphasis on the use of non-linear models. Thereafter we include a review of recent results on the topic of ANN. The relevance of ANN models for the statistical methods is considered using time series prediction problems. Finally we construct asymptotic prediction intervals for ANN and show how to use prediction intervals to choose the number of nodes in the ANN.

Détail

Classification : 62M10, 62M20, 62M45, 68T05, 82C32
Keywords: artificial neural network; non-linear time series model; prediction

Bibliographie

[1] Allende H., Galbiati J.: Robust test in time series model. J. Interamerican Statist. Inst. 1 (1996), 48. 35–79 | MR

[2] Allende H., Heiler S.: Recursive generalized M-estimates for autoregressive moving average models. J. Time Ser. Anal. 13 (1992), 1–18 | DOI | MR | Zbl

[3] Anderson B., Moore J.: Optimal Filtering. Prentice Hall, Englewood Cliffs, N.J. 1979 | Zbl

[4] Baxt W. G.: Use of an artificial neural network for data analysis in clinical decision marking: The diagnosis of acute coronary occlusion. Neural Computational 2 (1990), 480–489 | DOI

[5] Benitez J. M., Castro J. L., Requena J.: Are neural network black boxes? IEEE Trans. Neural Networks 8 (1997), 1156–1163 | DOI

[6] Beran J.: Statistics for Long-memory Processes. Chapman and Hall, London 1994 | MR | Zbl

[7] Bowerman B. L., O’Connell R. T.: Forecasting and time series: an applied approach. Third edition. Duxbury Press, 1993 | MR | Zbl

[8] Box G. E. P., Jenkins G. M., Reinsel G. C.: Time Series Analysis, Forecasting and Control. Third edition. Prentice Hall, Englewood Cliffs, N.J. 1994 | MR | Zbl

[9] Breiman L., Friedman J., Olshen, R., Stone C. J.: Classification and Regression Trees. Belmont, C. A. Wadsworth, 1984 | MR | Zbl

[10] Brockwell P. J., Davis R. A.: Time Series Theory and Methods. Springer Verlag, New York 1991 | MR | Zbl

[11] Brown R. G.: Smoothing, Forecasting and Prediction of Discrete Time Series. Prentice Hall, Englewood Cliffs, N.J. 1962 | Zbl

[12] Chatfield C.: Forecasting in the 1990s. Statistician 4 (1997), 46, 461–473

[13] Cheng B., Titterington D. M.: Neural networks: review from a statistical perspective. Statist. Sci. 1 (1994), 2–54 | DOI | MR

[14] Connor J. T., Martin R. D.: Recurrent neural networks and robust time series prediction. IEEE Trans. Neural Networks 2 (1994), 5, 240–253 | DOI

[15] Crato N., Ray B. K.: Model selection and forecasting for long-range dependent processes. Internat. J. Forecasting 15 (1996), 107–125 | DOI

[16] Fine T. L.: Feedforward Neural Network Methodology. Springer, New York 1999 | MR | Zbl

[17] Flury B., Riedwyl H.: Multivariate Statistics: A Practical Approach. Chapman Hall, London 1990

[18] Friedman J. H.: Multivariate adaptive regression spline. Ann. Statist. 19 (1991), 1–141 | DOI | MR

[19] Funahashi K. I.: On the approximate realization of continuous mappings by neural networks. Neural Networks 2 (1989), 183–192 | DOI

[20] Han J., Moraga, C., Sinne S.: Optimization of feedforward neural networks. Engrg. Appl. Artificial Intelligence 2 (1996), 9, 109–119 | DOI

[21] Hornik K., Stinchcombe, M., White H.: Multilayer feedforward networks are universal approximators. Neural Networks 2 (1989), 359–366 | DOI

[22] Hristev R. M.: Artificial Neural Networks. Preprint of a book obtained via Internet from the author, 1998

[23] Hutchinson J. M.: A Radial Basis Function Approach to Financial Time Series Analysis. Ph.D. Thesis. Massachusetts Institute of Technology, 1994 | MR

[24] Hwang J. T. G., Ding A. A.: Prediction for artificial neural networks. J. Amer. Statist. Assoc. 92 (1997), 438, 748–757 | MR

[25] Lin J. L., Granger C. W.: Forecasting from non-linear models in practice. Internat. J. Forecasting 13 (1994), 1–9 | DOI

[26] Lippmann R. P.: An introduction to computing with neural nets. IEEE ASSP Magazine (1997), 4–22

[27] Ljung G. M., Bax G. E. P.: On a measure of lack of fit in time series models. Biometrika 65 (1978), 297–303 | DOI

[28] McCullagh P., Nelder J. A.: Generalized Linear Models. Chapman Hall, London 1989 | MR | Zbl

[29] Meditch J. S.: Stochastic Optimal linear Estimation and Control. MacGraw–Hill, New York 1969 | Zbl

[30] Moody J. E., Utans J.: Architecture selection strategies for neural networks. In: Refenes A. P. N. Neural Networks in the Capital Markets, Wiley, New York 1995

[31] Moraga C.: Properties of parametric feedforward neural networks. In: XXIII Conferencia Latinoamericana de Informática, Valparaíso 1997, Vol. 2, pp. 861–870

[32] Pineda F. J.: Generalization of Backpropagation to recurrent and higher order networks. In: Proc. IEEE Conf. Neural Inform. Proc. Syst., 1987

[33] Poli R., Cagnoni S., Coppini, G., Walli G.: A neural network expert system for diagnosing and treating hypertension. Computer (1991), 64–71

[34] Referes A. P. N., Zapranis A. D.: Neural model identification, variable selection and model adequacy. J. Forecasting 18 (1999), 299–322 | DOI

[35] Reinsel G. C.: Elements of Multivariate Time Series Analysis. Springer Verlag, New York 1993 | MR | Zbl

[36] Ripley B. D.: Statistical aspects of neural networks. In: Networks and Chaos-Statistical and Probabilistic Aspect (O. E. Barndorf–Nielsen, J. L. Jensen, and W. S. Kendall, eds.), Chapman and Hall, London 1993 | MR | Zbl

[37] Sarle W. S.: Neural networks and statistical methods. In: Proc. of the 19th Anual SAS Users Group International Conference, 1994

[38] Smith J., Yadav S.: Forecasting cost incurred from unit differencing fractionally integrated processes. Internat. J. Forecasting 10 (1994), 507–514 | DOI

[39] Stern H. S.: Neural networks in applied statistics. Technometrics 38 (1996), 3, 205–214 | DOI | MR | Zbl

[40] Rao T. Subba: On the theory of bilinear models. J. Roy. Statist. Soc. Ser. B 43 (1981), 244–255 | MR

[41] Sussmann H. J.: Uniqueness of the weights for minimal feedforward nets with a given input-output map. Neural Networks 5 (1992), 589–593 | DOI

[42] Temme K. H., Heider, R., Moraga C.: Generalized neural networks for fuzzy modeling. In: Proc. Internat. Conference of European Society of Fuzzy Logic and Technology, EUSFLAT’99 Palma de Mallorca 1999

[43] Tong H.: Non-linear Time Series. Oxford University Press, Oxford 1990 | Zbl

[44] Vapnik V.: The Nature of Statistical Learning Theory. Springer Verlag, Berlin 1995 | MR | Zbl

[45] Vapnik V., Chervoneski A.: The necessary and sufficient conditions for consistency of the method of empirical risk minimization. Pattern Recognition Image Anal. 1 (1991), 284–305

[46] Waibel A., Hanazawa T., Hinton G., Shikano, K., Lang K. J.: Phoneme recognition using time delay neural networks. IEEE Trans. Acoust. Speech Signal Process. 37 (1989), 324–329 | DOI

[47] Wu F. Y., Yen K. K.: Application of neural network in regression analysis. In: Proc. 14th Annual Conference on Computers and Industrial Engineering, 1992