TECHCOMB

Volume 1 : Issue 1

Lagged Variables Selection for Fuzzy Time Series Models by Using Binary Particle Swarm Optimization

Authors : Erol Egrioglu , Bahadir Ozdemir

image

Abstract:

The selection of lagged variables is an important problem for fuzzy time series forecasting model. Many real-world time series need high order fuzzy time series models. In the literature, there is no method to select lagged variables in fuzzy time series models. In this study, a new fuzzy time series algorithm was proposed and also binary particle swarm optimization was used for selecting lagged variables. Fuzzy time series methods have three stages. These stages are fuzzification, determining of fuzzy relations and defuzzification stages, respectively. In the proposed method, fuzzy c-means method, min-max compositions and centralization methods were used in fuzzification, determining of fuzzy relations and defuzzification stages, respectively. Besides, it is aimed to solve high order fuzzy time series forecasting method by using fuzzy relation matrix in this study. The proposed method was applied to three stock exchange data time series. The results show that the forecasts of the proposed method are more accurate than forecasts of some other fuzzy time series methods in the literature.