應用存活分析法於運具移轉行為之研究

A Study on Mode Transfer Behavior Using Survival Analysis

莊君偉(Chun-wei Chuang)

指導教授 ： 吳健生

國立中央大學/工學院/土木工程學系/碩士(2011年)

摘要

在台灣由於私人交通工具的大量使用，造成空氣汙染、尖峰時段道路擁塞等問題，因此為解決私人交通工具大量使用的情況，政府近幾年來積極推動綠色運輸與鼓勵民眾改搭乘大眾交通工具，基於此相關議題，本研究以問卷設計私人交通工具使用者移轉大眾交通工具之影響，並利用存活分析法建構私人交通工具使用者運具移轉行為模式。
研究顯示，在已發生運具移轉之私人交通工具使用者中，平均使用私人交通工具約74個月，會發生運具移轉行為。進一步探討各解釋變數與使用者運具移轉之相關性，得知「油價上漲」、「使用天數」、「年齡」、「家庭人數」、「教育程度」、「職業」六項變數，再以log- log存活曲線檢定出部分變數不符合等比例模式假設，因此必須以分層Cox等比例危險模式進行配適，才能將不符合Cox等比例模式假設納入校估，最後依年齡分為兩層校估出五個顯著變數：「油價上漲」、「使用天數」、「家庭人數」、「教育程度」、「職業」與使用者發生運具移轉行為有顯著性。
進一步利用汽車移轉至大眾交通工具與機車移轉至大眾交通工之風險不同分別建構競爭風險模式，結果顯示「年齡」在影響汽機車移轉行為上皆具有顯著之變數，而原先使用汽車者比原先使用機車者還要容易發生運具移轉行為，另外「家到目的地距離」與「教育程度」則會影響機車移轉行為；「所得」會影響汽車移轉行為。

關鍵字

分層Cox等比例危險模式 ； 運具移轉 ； 相對風險模式 ； 競爭風險模式

並列摘要

Private mode use is a major cause of the emissions of pollution and peak of traffic congested in Taiwan. Therefore, To solution Private modes use, the government promotes to use green transportation and hope people give up private mode to use public mode.And this research investigates time internal value that private mode user transfer public mode behavior by using questionnaire. and it used survival analysis to model the sample.
The results have shown that the average survival time is 74 months which means that private mode user of have occurred mode transfer. Furthermore, the results have significant effects between the covariates and private mode user transfer public mode behavior. the result shows six covariates：「the oil price rising」、「days of use mode per week」、「age」、「number of members in his family」、「education」、「occupation」. And, use log minus log method to test Cox proportional hazard model. Finally, according to「age」, divide into two groups and construct each group to become the stratified Cox proportional hazard model. After the model testing and variable selection, there are five variables 「the oil price rising」、「days of use mode per week」、「number of members in his family」、「education」、「occupation」in model.
In this study use two different natures of risk-specific events – motorcycle user transfer mode behavior and car user transfer mode behavior. The result shows「age」covariates affect the motorcycle user transfer mode behavior and the car user transfer mode behavior,and car user is easier to transfer mode than motorcycle user；besides「home to destination of distance」and「education」affect motorcycle user transfer mode behavior；「income」covariates also affect car user transfer mode behavior.

並列關鍵字

competing-risks model ； relative risk ； stratified Cox proportional-hazards model ； mode transfer

參考文獻

1. Villcasim N. J. and D. C. Jain（1991）,“Modeling Purchase-Timing and Brand-Switching Behavior Incorperating Explanatory Variables and Unobserved Heteroheneity”,Journal of Marketing Research,Vol.XXVIII, pp29-41.

2. Gonul F.and K.Srinivasan（1993）,“Consumer Purchase Behavior in a Frequently Bought Product Category:Estimation Issues and Managerial Insights from a Hazard Function Model with Heterogeneity”,Journal of American Statistical Association,Vol88,No.424,pp1219-1227.

3. Mannering, F.（1993）,“Male/female driver characteristics and accident risk: some new evidence.” Accident Analysis and Prevention Vol.25,pp.77–84.

4. Chandra, R. B., “A Generalized Multiple Durations Proportional Hazard Model with an Application to Activity Behavior During Evening Work-to-Home Commute”,Transportation Research Part B,Vol.30,No. 6,1996,pp.465-480.

5. De Jong G.（1996）,“A Disaggregate Model System of Vehicle Holding Duration, Type Choice and Use”,Transportation Research Part B, Vol.30,No.4,pp263-276.