Abstract: The problem of prediction of British Petroleum Corp. stock prices and the Dow Jones Industrial Average stock quote is considered. For the prediction data stock quote of the largest oil companies at the stock exchange NYSE were used as input data. The obtained experimental results of prediction using FGMDH were compared with the classical GMDH and cascade neo-fuzzy neural networks. For the classical and fuzzy GMDH four classes of functions- linear, quadratic, Fourier polynomial and Chebyshev polynomial were used, and the variation in the form of membership function, the size of learning sample and freedom of choice with the developed software were performed. Experimental results of forecasting at NYSE are presented enabling to estimate efficiency of different forecasting methods and to choose the most proper method.

Keywords: fuzzy group method of data handling, stock exchange, stock prices forecasting, cascade neo-fuzzy neural networks.

Link:

НЕЧЕТКИЙ МЕТОД ИНДУКТИВНОГО МОДЕЛИРОВАНИЯ В ЗАДАЧАХ ПРОГНОЗИРОВАНИЯ НА ФОНДОВЫХ РЫНКАХ

Юрий Зайченко

http://www.foibg.com/ijima/vol01/ijima01-4-p01.pdf