A new piecewise fuzzy exponential smoothing model based on some change-points

Research highlights? Limited time series data with some change points is a difficult task in forecasting. ? We examine the change-points in the collected data and use them to fit a fuzzy exponential smoothing model to increase quality of forecasting. ? Under the circumstance of some fluctuated points in the collected data, a piecewise fuzzy exponential smoothing model can be easily applied. In our previous studies we proposed the fuzzy exponential smoothing model for extrapolation under a vague system with limited data. However, some change-points in the collected data always generate to enlarge a forecasting interval which provides the decision maker with little information to make some decisions. Therefore, in this study we defined the change-points and the piece forecasting intervals for deriving the piecewise fuzzy exponential smoothing interval, and we can effectively determine the future trends for decision. Finally, an illustrated example has been used to verify the effectiveness and confirm the potential benefits of the proposed model with a smaller and piecewise forecasting interval.