查詢結果分析
來源資料
頁籤選單縮合
題名 | 臺灣地區小汽車成長預測模式之建立 |
---|---|
作 者 | 藍武王; 林麗玉; | 書刊名 | 中國統計學報 |
卷期 | 29:1 1991.03[民80.03] |
頁次 | 頁49-75 |
分類號 | 484.3 |
關鍵詞 | 小汽車; 成長; 預測; 臺灣地區; 模式; |
語文 | 中文(Chinese) |
中文摘要 | 小汽車成長之預測不僅可作為運輸部門重要施政及管制措施之參據,對於國家總體能源政策、土地使用計劃、汽車工業發展等政府部門亦很重要。本文以民國41年至76年台灣地區小汽車總車輛數為因變變,影響小汽車成長之社經因素自變數,建立多元迴歸模式。模式構建過程包括模式型態認定、自變數之選擇相關分析、模式參數校估、殘差變異同質性檢定、自我相關檢定等,最後得到最佳線性不偏性(BULE)之預測模式。本預測模式僅包括四項最重要之自變數:公路高級路面里程數、小汽車購置能力(以國民生產毛額除以小汽車價格表示)、小汽車進口關稅及購車自備停車位之虛擬變數,其中自變數與小汽車總數多呈非線性關係。此外,本文亦利用逐步迴歸分析加上先驗知識判斷,建立逐步迴歸模式。經輸入七十七、七十入年自變數資料由模式求得預測值,再與實際值及國內其他小汽車預測模式之預測結果相比較,發現本文所建立之二模式其預測準確度較佳,但仍有低估現象,顯示近兩年來小汽車正處於急速成長階段,除本文構建預測棋式所含蓋之說明變量,可能尚有短期簡易計量之因素影響汽車之急迫成長,值得進一步研究。 |
英文摘要 | Precisely forecasting the growth passcnger cars usually plays an important role on policy planning in the t1'ansportation sector. Other public sectors may also need such forecasted data to develop related policies such as national energy poliey land-use policy, and vehicle industry development. This paper constructs a multple regression model to forecast the growth of passenger cars in Taiwan Area. Time series data frorn 1952 through 1987 including the nurnber of passenger cars and the socio-economic factors affecting the ownership of passenger cars are used to calibrate this model. Details of the model construction procedures are described, including priori causal relationship analysis, function form identification. correlation analysis, model estimation, homoscedasticity test, and autocorrelation test. A BLUE forecasting model is finally obtained in which four explanatory variables-length of paved highway, car purchasing ability (GNP/car price), car tariffs. and expectance of requiring private parking space are included. It was found that the number of passenger cars and each of the explanatory variables are nonlinearly related. Furthemore, anther forecast model is built based on the stepwise regression technique in conjunction with priori knowledge. In order to justify the forecasting power. the true values of car number of 1988 and 1989 are used to compare with the theoretical ones forecasted by these two models and by other previously developed models. It shows that these two models are much better than others in terms of the forecasted erros. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。