Fuzzy Time-Series Analysis using Class Value

Abstract

An interval type AR model uses stationary time-series transformed into interval fuzzy numbers. The high forecasting accuracy of the model using interval type stationary time-series has been confirmed in previous case studies. On the other hand, forecasting accuracy is not always required in time-series forecasting. In this case, a prediction model should be easy to handle and show some trend. It is confirmed that the fuzzy time-series model satisfies this requirement when stationary time-series classified into classes are used. For this reason, this study reports the results of a confirmed study.