Fuzzy forecasting based on linear combinations of independent variables, subtractive clustering algorithm and artificial bee colony algorithm

Abstract

In this paper, we propose a new clustering-based fuzzy time series (FTS) forecasting method based on linear combinations of independent variables, the subtractive clustering algorithm and the artificial bee colony (ABC) algorithm. The subtractive clustering algorithm is used to automatically generate clusters of historical training data and get the cluster center of each cluster. The ABC algorithm is applied to obtain the optimal neighborhood radiuses of the subtractive clustering algorithm to get the optimal cluster center of each cluster of the historical training data. Based on the obtained cluster center of each cluster, the weighted contribution of each cluster with respect to each historical training datum is calculated. Finally, the proposed method constructs the linear combinations of independent variables of the historical training data using this weighted contribution to forecast the historical testing data. The proposed method gets higher forecasting accuracy rates than the existing methods for forecasting the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX), the enrollments of the University of Alabama and the daily percentage of CO2.