The developing of fuzzy system for multiple time series forecasting with generated rule bases and optimized consequence part

The paper aims to build and implement a predictive model called a fuzzy system. The fuzzy rule bases component is generated by using the input-output data pairs. Its consequence part is optimized by using ordinary least squares. The initial structure model is needed to create the input-output data pairs based on the multiple time series. The rule bases are generated by using table lookup schema in which each input-output pairs has a contribution as a candidate rule. The obtaining rule base is modified to be an efficient one by optimizing its consequence part. As a case study is used, the 2-time series assumed which they have a causality effect. The data are the soybean price of both domestic and abroad. The developed fuzzy system is used in the forecasting of the domestic soybean price. The fuzzy system's performance is very satisfying, assessed according to the R-squared and mean squared error of criteria.