Forecasting Seasonal Fuzzy Time Series via Periodical Analysis Approach

Fuzzy time series forecasting methods has got more and more attention in recent years since they have a good capability of forecasting real-world time series which contains uncertainty. There have been various fuzzy time series forecasting methods in the literature. On the other hand, just a few ones have been proposed to forecast seasonal time series. When a seasonal time series is forecasted, there are few options to forecast such time series. If periodical analysis approach is used, it is possible to use any nonseasonal fuzzy time series forecasting approach in order to forecast seasonal time series. In this study, periodical analysis approach is introduced for fuzzy time series methods. By utilizing two non-seasonal fuzzy time series approach proposed by Tsaur (2011) and Chen (1996), a well-known real world seasonal time series is forecasted with periodical analysis approach. For comparison, the real world time series is also forecasted by SARIMA models. It is seen from the obtained forecasting results that periodical analysis approach produces very good forecasts for a well-know real world seasonal time series.