透過您的圖書館登入
IP:3.144.38.130
  • 期刊

Comparison of Alternative Tourism Forecasting Models

觀光休憩預測模型之比較

摘要


本研究著重在比較數種觀光遊憩預測模型預測美國密西根州內各區域觀光遊憩量的能力。密西根州被劃分為九大區域。這九大區域包括密西根州本身和八個區域。以1974至1984年的密西根州月觀光遊憩職業人口(Tourism-related Employment)時間序列為數據。使用相同的分析和預測程序為每個區域發展出四種不同的預測模型並評估和比較各模型的預測精確程度。這四種模型包括結構式迴歸模型,時間序列式迴歸模型,結構式轉換函數模型(Structural Transfer Function model),和時間序列式轉換函數模型(Time-series Transfer Function Model)。雖然轉換函數模型看起來比多元迴歸模型要複雜的多,但我們可以使用TSP統計套裝軟體且不須多少額外的時間和努力便可得出轉換函數模型。所有的預測模型都能精確地預測過去和未來的觀光遊憩職業人口。大多數的結構式迴歸模型具有比時間序列式迴歸模型較小的預測誤差。迴歸模型的預測精確度可因運用轉換函數方法於其迴歸剩餘上而顯著地提高。對北部三個在經濟上依賴旅遊事業的區域而言,結構式轉換函數模型比時間序列式轉換函數模型表現得要好。但時間序列式轉換函數模型在其他區域中表現得要比結構式轉換函數模型要好一些。結構式和時間序列式轉換函數模型間的選擇端視於研究的目的,合適的解釋變數的可得性,以及預測解釋變數的能力而定。

關鍵字

預測 迴歸 轉換函數

並列摘要


This paper investigates the ability of alternative models to forecast levels of tourism activity in various regions of Michigan. Four distinct trend models were developed and evaluated for each of eight sub-regions of the state using monthly tourism-related employment data for the period 1974-1984. A structural and a pure time series regression model were tested along with corresponding transfer function models. Most of the structural regression models have smaller forecast errors than the time series regression models. Forecast accuracy of the regression models was improved by applying transfer function techniques to each model's residuals. The structural transfer function models have smaller forecast errors for tourism - dependent regions of Michigan while the time series transfer function models perform better for other regions of the state. The choice betweens tructural and time series transfer function models depends largely on the purpose of the study, the availability of suitable explanatory variables, and the ability to forecast the explanatory variables.

並列關鍵字

Forecast regression transter function

被引用紀錄


王心怡(2004)。臺灣地區遊憩需求預測之分析〔碩士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2004.01882

延伸閱讀