Authors: Yoshiyuki Yabuuchi; Takayuki Kawaura
Addresses: Faculty of Economics, Shimonoseki City University, 2-1-1 Daigaku-cho, Shimonoseki, Yamaguchi 751-8510, Japan ' Department of Mathematics, Kansai Medical University, 2-5-1 Shin-machi, Hirakata, Osaka, 573-1010, Japan
Abstract: Economic systems are complex because they are influenced by many factors such as policy impacts, human behaviour, and consciousness. When a system includes these factors, a fuzzy system approach plays a pivotal role in its analysis. A fuzzy autocorrelation model proposed by Yabuuchi et al. uses the concept of soft computing for the Box-Jenkins model. In a case study using a fuzzy autocorrelation model, fuzzy random variables were used for the fuzzy autocorrelation model, because of the unnatural behaviour of the predicted value. The results revealed an improvement in the predictive accuracy of the model. In this study, we validate the feasibility of the proposed approach by using the results of an analysis of the Japanese national consumer price index. In this analysis, we show that unnatural behaviour and ambiguities of the proposed model are improved by using fuzzy confidence intervals.
Keywords: fuzzy time series; Box-Jenkins model; autocorrelation; fuzzy numbers; fuzzy random variables; fuzzy intervals; fuzzy confidence intervals; consumer price index; Japan; economic analysis; consumer price index.
International Journal of Advanced Mechatronic Systems, 2016 Vol.7 No.1, pp.46 - 60
Accepted: 30 Mar 2016
Published online: 04 Oct 2016 *