台灣高鐵ARIMA法時間序列迴歸法分解法Taiwan High Speed RailARIMATime series regressionDecomposition
在科技越來越進步的趨勢下，各種交通工具的出現，提供民眾許多選擇。此外對於大眾運輸服務水準以及乘坐時間要求也逐漸提高，所以選擇搭乘高鐵的人逐年遞增，我們想知道未來高鐵成長趨勢會如何，因此希望透過課程學習的統計預測方法，找出一個最佳模型來預測未來一年高鐵每月進站人數。並期待這份報告能夠提供給高鐵單位作為參考。本報告主要研究2007年1月至2018年12月臺灣高鐵每月進站旅客人數，資料來源為台灣高鐵網站上提供的數據。此研究將透過三個分析方法:ARIMA法、時間序列迴歸法和分解法進行分析，並建立了四個模型，再利用三種評估指標: MAD(mean absolute deviation)、MSE(mean square error)及MAPE(mean absolute percentage error)作為選出最佳模型的依據，最後分析結果顯示，乘法分解模型為最佳模型。Abstract
With the increasing progress of science and technology, the emergence of various modes of transport provides many choices for the public. Besides, the level of public transport services and travel time requirements are gradually increasing, so the number of people choosing to take high-speed rail is increasing year by year. We would like to know the future growth trend of high-speed rail, so we hope to learn through the course of statistical forecasting methods, to find out the best model to predict the number of high-speed rail monthly inbound in the coming year. It is also expected that this report can be provided to high-speed rail units for reference. This report focuses on the number of monthly passengers entering Taiwan's high-speed rail from January 2007 to December 2018, based on data available on the Taiwan high-speed rail website. This study will be analyzed by three analytical methods: ARIMA, time series regression, and decomposition, and four models will be established, using three evaluation indicators: MAD (mean absolute deviation), MSE (mean square error), and MAPE (mean absolute percentage error), as the basis for selecting the best model, and finally the results show that the multiplication decomposition model is the best.