Tourism Demand Forecasting Using a Novel High-precision Fuzzy Time Series Model

Fuzzy time series model has been developed to either improve forecasting accuracy or reduce computation time, whereas a residul analysis in order to improve its forecasting performance is still lack of consideration. In this paper, we propose a novel Fourier method to revise the analysis of residual terms, and then we illustrate it to forecast the Japanese tourists visiting in Taiwan per year. The forecasting results show that our proposed method can derive the best forecasting performance as well as the smallest forecasting error of MAPE in the training sets; in the testing sets, the proposed model is also better to fit the future trend than some forecasting models.