國內航空載客現況與預測

Abstract

本報告主要研究每年國內航空載客量，期望藉由過去資料推測未來趨勢。資料來源為主計處總體統計資料庫，資料時間從2002年1月至2016年3月，共171筆資料，其中我們保留最後十二筆資料作為預測的對照依據。時間序列分析方法使用ARIMA分析法、時間序列迴歸法、分解法以及指數平滑法，配適模型並進行預測。本報告採用四種評估準則: MSE (Mean Squared Error)、MAE (Mean Absolute Error)、MAPE (Mean Absolute Percentage Error)，以及MPE (Mean Percentage Error)，將十二筆預測結果與最後十二筆實際資料作比較，檢驗預估是否準確。結果發現時間序列迴歸法為最佳時間預測方法。

The objective of this study is to forecast the monthly number of passengers of Taiwan’s domestic flights. We predict 12-step-ahead numbers of passengers based on four forecasting models. Our data source is the macroeconomic database from National Statistics. The duration of data is from January 2002 to March 2016. There are 171 data during the period, and we hold out the last 12 observations for the out-of-sample forecasts. The forecasting methods include ARIMA (Autoregressive Integrated Moving Average) model, time series regression, decomposition methods, and exponential smoothing method. We use four evaluation criteria to examine the accuracy of the forecast, which are MSE (Mean Squared Error), MAE (Mean Absolute Error), MAPE (Mean Absolute Percentage Error), and MPE (Mean Percentage Error). It turns out that the time series regression is the best forecasting model based on the four criteria.