臺灣國際觀光旅館住用率研究

Abstract

本文主要為研究台灣國際觀光旅館住用率的資料預測，數據資料來源主要從自「中華民國交通部觀光局」網站所發行的每月統計營運報告中，擷取台灣國際觀光旅館的住客率資料，時間從自西元1999年1月起至2012年3月，共159筆月資料。

本研究主要使用統計預測中常用的四種預測方法，分別為ARIMA模型分析法，時間序列迴歸法、時間序列分解法以及指數平滑法。於每一分析預測方法，透過利用配適模型以及對其作自我相關的檢定，並且使用模型所預測的數據與真實值做比對，再利用預測評價指標判斷最合適之預測模型。

最終，本報告得到最合適之用於預測台灣國際觀光旅館之住客率的模型是為時間序列迴歸法，其精準預測住客率之未來十二筆數據值。

Abstract The principal aim of this paper is to research the occupancy rate of international hotel in Taiwan. All the data and related information have been retrieved from the websites of the Tourism Bureau, Ministry of Transportation of Republic of China including the monthly statistics operation reports . We only gathered t the occupancy rate of international hotel in Taiwan between January 1999 to March 2012. These data contain 159 monthly data. In this research four common types of forecast measurement were used the ARIMA Model Analysis Method, the Time Series Regression Method, the Exponential Smoothing Method and the Decomposition Method. Autocorrelation of each model was taken and the their respective predictions compared to the real values. Furthermore, we used the forecast evaluation instruction to determine the most appropriate forecast model. As a result, the most appropriate model to predict the occupancy rate of international hotel in Taiwan is the Time Series Regression Method. It accurately predicted the future occupancy rate twelve times.