Functional coefficient autoregressive time series modeling based on fuzzy inference method

Abstract

Functional coefficient autoregressive model is a class of useful nonlinear time series models. A new approach based on fuzzy inference modeling method for the estimation of the coefficient functions is proposed in this paper. A corresponding algorithm which is easy to be realized is designed. Besides, two typical simulation examples are studied, the estimated errors of which are minor than other models results. And the performance of forecasts on the proposed models has high degree of accuracy, which illustrates the validity of the method further.