SEASONAL FORECASTING OF A DECOMPOSED FUZZY EXPONENTIAL SMOOTHING MODEL USING GREY ESTIMATED VALUES

This paper proposes a decomposed fuzzy exponential smoothing model to analyze the seasonal time series data in which the secular trend and seasonal effects are transferred. Using the transferred data and the estimated grey value, the fuzzy exponential smoothing equation is solved, resulting in a smaller forecasting error. Then, the decomposed forecasting values are traced back by multiplying the seasonal index and trend values to obtain the seasonal forecasting values. Three examples are provided to illustrate the proposed model. In the training sets, the forecasting errors of the proposed model are better than Holt-Winter's model and the statistically decomposed method. In the test sets, the proposed model is also a better fit for the future trend.