臺灣旅遊需求預測-轉移函數模型的應用

Abstract

隨著全球旅遊市場快速發展，臺灣旅遊產業的發展日益興盛，旅遊觀光產業對於臺灣經濟與社會帶來了可觀的貢獻，重要性不容小覷。本研究以來臺旅客人次為研究對象，並結合片段線性模型與時間序列分析法建構轉移函數模型，針對臺灣旅遊需求進行預測。以評價模型預測能力指標－平均絕對百分比誤差(Mean Absolute Percentage Error, MAPE)及均方根誤差(Root Mean Square Error, RMSE)評估預測精確度，並進一步加入簡算法(Naive)作為指標模型，輔以Diebold-Mariano檢定進行各模型預測能力之比較。實證結果顯示片段線性模型和轉移函數模型較簡算法對於臺灣旅遊需求之預測準確，兩者皆具有相當優異的預測能力，期望研究結果能為相關研究領域進行多一層的探討。

Taiwan’s tourism industry grows with significant contribution to the society as that of the whole world develops. This research combines the piecewise linear model and the time series analysis method to build a transfer function model for forecasting the demand for Taiwan tourism based on the monthly tourist arrivals. In addition, we use the mean absolute percentage error (MAPE) and root mean square error (RMSE) to assess the precision of the forecasting models.Finally, in order to compare the out-of-sample forecasting accuracy between different models, Naive method is added as a benchmark model and evaluates forecasting performance between two models by using the Diebold-Mariano test. The result turns out to be that piecewise linear model and transfer function model predict Taiwan tourism demand precisely and they are significantly outperform Naive method for the out-of-sample forecasting period. Hope this research can make a contribution to the relevant research fields.