Forecasting of Time Series Using Fuzzy Logic and Particle Swarm Optimization Algorithm

During the last few decades, fuzzy time series have been developed so as to better the exactness of forecasting. In this study, we suggest a hybrid algorithm of fuzzy time series and particle swarm optimization (PSO) algorithm to solve the forecasting problem. Such algorithm is considered a very effective and a recent method. It is inspired by 'birds' flight and communication behaviors. The used algorithm assigns the length of each interval in the universe of discourse, and the degree of membership values, and updates weights. The selected data sets are used to clarify the suggested method, and then compare the forecasting exactness with another method, that used a hybrid model based on the statistical model (ARIMA) and Artificial Neural Network (ANN). The results show that the suggested algorithm is more exact in forecasting time series, and can compete well with other methods.