觀光業帶來的商機，給民眾帶來更多的財富，觀光活動也可讓民眾在工作之餘，從事休閒育樂活動，平衡忙碌的生活。觀光產業還可以與自然資源、商務活動結合，進一步透過觀光交流，展現一國的人文內涵、經濟實力及基礎建設現代化程度，對改善整體環境、提升國家文化素質均助益匪淺。 日本一直都是台灣在旅遊市場中最重要客源，透過正確的分析和完善的規劃與管理才能使旅遊市場的供需達到均衡。因此，必須準確地預測來台觀光旅客需求，才以掌握旅遊市場狀況與發展，以進一步規劃各種軟硬體設施的投資，例如：大規模飯店興建、遊覽車購置、導遊培訓…等；反之，不適當的評估或是不精確的預測，將導致觀光資源不敷使用或閒置浪費。 由於統計迴歸模式必需收集完整的變數以建構預測模式，當收集的資料受到限制，時間序列資料存在語意值或是資料量少於50筆時，統計時間序列通常會因為無法得到較小的誤差而失效。為了處理這樣的問題，本研究首先提出採用前期資料的適應性模糊時間序列模式進行分析，結果顯示其績效不佳，且模式無法行進外差的預測。為改善此項缺點，本研究再提出一個具殘差修正的模糊時間序列模式，來處理外差預測的問題，預測結果顯示，無論是內差或外差，其MAPE與RMSE皆相當準確。 Over the past few decades, the tourism industry has been grown very fast. Because of the tourism activity may for the country creation traveling income, plan and the management sightseeing resources because of the forecasting result. Thus, it is very important for planning for potential tourism demand and improving the tourism infrastructure, since accurate forecasting of tourist arrivals. Japan has been the most important source that Taiwan travels all the time. But the international exchange is frequent day by day, the competition of the tour undertaking is fiercer and fiercer, only depend on correct decision and planning and management of perfection. There are many method can forecast, but when the collected are not enough to model regression model or time series model, or there exist fuzzy time series data, the statistical quantitative methods are usually failure to have smaller forecasting error. In order to provide a much more flexible examination for managing smaller data set or fuzzy data. In this study, we proposed an adaptive fuzzy time series model for forecasting tourism demand for Japanese to Taiwan. But it can’t forecast accurately, and it can’t forecast about untrained data, so we proposed a new method which combined Fourier series with fuzzy time series for forecasting Japanese tourism demand for Taiwan, and obtained very small forecasting error MAPE and RMSE.
