Prediction of time series based on a universal measure and decision trees
Abstract
In this paper, we propose and develop two methods for time series forecasting based on the methods of data compression. Theoretical justification of the methods described is presented, as well as the ways of using the prediction methods to the real series forecasting. In addition, the results of experimental studies of two methods are considered illustrated by real economic series forecasting such as indexes of industrial and consumer prices and exchange rates. The effectiveness of these techniques and methods for selecting effective parameters of these methods are also investigated.
About the Authors
A. S. Lysiak
СибГУТИ; НГУ
Russian Federation
Russian Federation
B. J. Ryabko
СибГУТИ
Russian Federation
Russian Federation
References
1. Ahmed N. An empirical comparison of machine learning models for time series forecasting // Econometric Reviews. 2010,Vol. 29, Issue 5-6. P. 594-621.
2. Palit A. K., Popovic D. Computational Intelligence in Time Series Forecasting: Theory and Engineering Applications (Advances in Industrial Control). Springer-Verlag New York: Secaucus, NJ, USA, 2005.
3. Zhang G., Patuwo B. E., Michael Y. H. Forecasting with articial neural networks:: The state of the art // International Journal of Forecasting. 1998. Vol. 14, Issue 1. P. 35--62.
4. Приставка П.А. Экспериментальное исследование метода прогнозирования, основанного на универсальных кодах // Вестник СибГУТИ, 2010. №4, C. 26--35.
5. Cheng H. et al. Multistep-ahead time series prediction // Lecture Notes in Computer Science. 2006. V. 3918. P. 765--774.
6. Bontempi G. Local Learning Techniques for Modeling, Prediction and Control. Ph.d., IRIDIA-Universit de Libre de Bruxelles, BELGIUM, 1999.
7. B. Ryabko. Compression-Based Methods for Nonparametric Prediction and Estimation of Some Characteristics of Time Series. // IEEE Transactions on Information Theory, Vol. 55, No. 9, 2009.P. 4309--4315.
8. Рябко Б. Я. Дважды универсальное кодирование // Проблемы передачи информации. 1984. Т. 20, № 3. С. 24--28
9. Рябко Б., Монарёв В. Экспериментальное исследование методов прогнозирования, основанных на алгоритмах сжатия данных // Проблемы передачи информации. 2005. C. 65--69.
10. Nevill-Manning C.G., Witten I.H., Paynter G.W. Lexically-GeneratedSubjectHierarchiesforBrowsingLargeCollections // International Journalof Digital Libraries. 1999. Vol. 2, Issue 3. P. 111--123.
11. Nevill-Manning C.G., Witten I.H.Identifying Hierarchical Structure in Sequences: A linear-time algorithm // Journal of Artificial Intelligence Research. 1997. Vol. 7. P. 67--82.
12. Poskitt D.S., Tremayne A.R. The selection and use of linear and bilinear time series models // International Journal of Forecasting. 1986. Vol. 2, Issue 1. P. 101--114
13. Tong H. Non-linear Time Series: A Dynamical System Approach. Oxford University Press, 1990.
14. Tong H. Threshold models in Nonlinear Time Series Analysis. Springer Verlag, Berlin, 1983.
15. Tong H., Lim K. S. Threshold autoregression, limit cycles and cyclical data //. Journal of the Royal Statistical. Series B (Methodological). 1980. Vol. 42, Issue 3. P. 245--292.
16. Engle R. Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom // Econometrica. 1982. Vol. 50. Issue 4. P. 987--1007.
17. Clements M.P. et al. Forecasting economic and financial time-series with non-linear models // International Journal of Forecasting. 2004. Vol. 20.Issue 2. P. 169--183
18. Рябко Б.Я. Прогнозирование случайных поседовательностей и универсальное кодирование. // Проблемы передачи информации. 1988. №24. C.3--14.
19. Krichevsky R. Universal Compression and Retrival. KluverAcademicPublishers, 1993.
Review
For citations:
Lysiak A.S., Ryabko B.J. Prediction of time series based on a universal measure and decision trees. The Herald of the Siberian State University of Telecommunications and Information Science. 2014;(2):57-71. (In Russ.)
Views:
221
Email this article 
