A New Fuzzy Time Series Method Based On Artificial Bee Colony Algorithm

Traditional forecasting methods need strict assumptions such as normality and linearity. It is very difficult to satisfy these assumptions for real-world time series. Many realworld time series can be easily analyzed by using fuzzy time series methods since fuzzy time series methods do not require any strict assumptions. Therefore, fuzzy time series approaches have been getting more and more attractive in recent years. Artificial intelligent techniques have been employed for different aims in fuzzy time series methods. In the literature, some intelligent techniques such as fuzzy c-means, genetic algorithm and particle swarm optimization have been commonly used for fuzzification of time series. In this study, a new fuzzy time series method in which artificial bee colony algorithm is utilized for fuzzification is firstly proposed. The proposed method is applied for different data sets of Istanbul Stock Exchange 100 Index. The obtained results show that the proposed method outperforms some other fuzzy time series methods avaliable in the literature.