Abstract
This paper aims to discuss the results and conclusions of an extensive comparative study on the forecasting performance between two different techniques: a genetic expert system in which a genetic algorithm carries out the identification stage embraced in the three- phase Box&Jenkins univariate methodology; and a connectionist approach. At the heart of the former, an expert system rules the identification-estimation-diagnostic checking cyclical process to end up with the predictions provided by the SARIMA model which best fits the data. We will present the connectionist approach as technically equivalent to the latter process and due to its, alas, lack of any conclusive existent algorithm able to identify both the optimal model and architecture for a given problem, the three most common models presently at use and 20 different architectures for each model will be examined. It seems natural that if a comparison is to be made in order to provide a straight answer as to whether or not a connectionist approach outperforms the univariate Box&Jenkins methodology, the benchmark should clearly be the set of time series analysed in the work ‘Time Series Analysis. Forecasting and Control’ by G. E. Box and G. M. Jenkins. Series BJA through to BJG give a total of 1200 plus measures to evaluate and compare the predictive power for different models, architectures, prediction horizons and pre-processing transformations.
Similar content being viewed by others
References
Yule GU. On a method of investigating periodicities in disturbed series. Phil Trans 1927; A226: 267
Box GEP, Jenkins GM. Time Series Analysis-Forecasting and Control, Holden-Day, San Francisco, 1976
Valls M. Identificació Automàtica de Sèries Temporals, PhD Thesis, UPC Barcelona, 1983
Akaike H. A new look at the statistical model identification. IEEE Trans Auto Control 1974; AC-19: 716–723
Burg JP. Maximum Entropy Espectral Analysis, PhD Thesis, Stanford University, 1975
Goldberg DE. Genetic Algorithms in Search: Optimization and Machine Learning. Addison Wesley, CA, 1989
Tsay RS, Tiao GC. Consistent estimates of autoregressive parameters and extended sample autocorrelation function for stationarity and nonstationarity ARMA models. J Am Statist Assoc 1984; 79: 84–96
Pankratz A. Forecasting with Univariate Box-Jenkins Models. John Wiley & Sons, New York, 1983
Granger C, Anderson T. An Introduction to Bilinear Time Series Models. Vandenhoeck and Ruprecht, Gottingen, 1978
Tong H. Threshold Models in Non-linear Time Series Analysis: Lecture Notes in Statistics. Springer-Verlag, Berlin, 1983
Hertz J, Krogh A, Palmer R. Introduction to the theory of neural computation. Addison-Wesley, CA, 1991
Freeman J, Skapura D. Neural Networks Algorithms, Application. Addison-Wesley, CA, 1991
McClelland J, Rumelhart D. Parallel distributed processing. MIT Press, MA 1988
Refenes A. Constructive learning and its application to currency exchange rate forecasting, 1990
Gallant S. Three constructive algorithms for neural learning. 8th Ann Conf Cog Sci Soc, 1986
Pelillo M, Fanelli M. A method of pruning layered Feddforward NNs. Proc IWANN '93, 1993
Karnin E. A simple procedure for pruning BP trained NNs. IEEE Trans Neurol Networks 1990; V1: 325–333
Eberhart R, Dobbins R. Neural Network PC tools a practical guide. Academic Press, CA, 1990
Cybenko G. Continuous valued Nns with 2 hidden layers are sufficient. Technical report, Tufts University, MA, 1988
Cybenko G. Approximation by superpositions of a sigmoidal function. Math Control, Signals Syst 1989; 303–313
Tang Z, Almeida C, Fishwick D. Time series forecasting using NNs vs. B&J methodology. Simulations 1991
Weigend A, Huberman B,et al. Predicting the future; a connectionist approach. Intl J Neural Syst. 1990; V1: 193–209
Varfis A, Versino C. NNs for economic time series forecasting. Proc ICANN'90, 1990
Groot C, Würtz D. Analysis of univariate time series with connectionist nets. Proc. ICANN'90, 1990
Baestens D, Bergh W,et al. Estimating tax inflows at a public institution. Proc. NN Capital Markets, 1993
Tong H, Lim K. Threshold Autoregression, limit cycles and cyclical data. RSS 1980; B42: 245
Sharda R, Patil R. Connectionist approach to time series prediction: an empirical test. J Intell Manuf 1992; 3: 317–323
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Blake, J., Francino, P., Catot, J.M. et al. A comparative study for forecasting using neural networks vs genetically identified Box&Jenkins models. Neural Comput & Applic 3, 139–148 (1995). https://doi.org/10.1007/BF01414075
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01414075