'''Sample division with sliding interval as the method of accuracy increase for time series forecast'''. Nina Kondrashova, Andriy Pavlov, Yaroslav Pavlov

'''Abstract'''. ''The forecast of heteroskedastic time series is considered at the example of data about the change of world price of indicative petroleum sorts of Brent and Urals. Two methods of forecast models construction are considered as the sum of trend and the remainder model and also with the help of the first differences model. The algorithm of inductive modeling for sliding window with optimization of division (AMSWOD) is offered, which uses the two-stage division in the "sliding window" of variable length. The window length is determined from the condition of accuracy maximization at the examinations subsamples. A comparison of the forecast obtained AMSWOD is made with the forecast obtained by the well known methods.'' 

'''Keywords'''. Forecast, heteroskedastic time series, sliding window, two-stage division.

'''References'''.
[[br]]1. Ivakhnenko A. G.: The Group Method of Data Handling-A rival of the Method of Stochastic Approximation, Avtomatika, 1968, no.3 (In Russian).
[[br]]2. G. Box, Gw. M. Jenkins, Gr. Reinsel : Time Series Analysis: Forecasting & Control (3rd Edition), Prentice Hall, 1994, 592p. 
[[br]]3. Kantorovich G.G.: Time Series Analysis. Economic magazine of High school of economics, 7, no.1, 79-103, 2003. (In Russian)
[[br]]4. Sheludko O. I.: The GMDH algorithm with complete orthogonalized description for models synthesis using the results of the design of experiments. Avtomatica, 1974, no.5, pp.32-41. (In Russian). 
[[br]]5. Stepashko V.S., Kondrashova N.V.: Research of methods of sample division in the GMDH algorithms, Works of I International Conference on Inductive Modeling, Lviv, 1, P.1, 90-94, 2002. (in Russian).
[[br]]6. Vanin V.V., Pavlov A.V.: Development and application of self-organization algorithms for modeling of complex processes and objects which are represented by the point former, Proceedings of Tavria State agrotechnical academy. P.4, vol.24, Melitopol, 2004, pp.51-56. (In Ukrainian).
[[br]]7. Kondrashova N.V.: Influence of method of sample division in GMDH algorithms on accuracy of forecasting // Control Systems and Computers, 2003, no.2, pp.128-133 (In Russian).
[[br]]8. http://trade.tatcenter.ru