交通运输系统工程与信息 ›› 2010, Vol. 10 ›› Issue (6): 128-132 .

• 系统工程理论与方法 • 上一篇    下一篇

基于拟合优度度量的交通方式选择决策树结构研究

温惠英*;曾文创   

  1. 华南理工大学 土木与交通学院,广州 510640
  • 收稿日期:2010-03-31 修回日期:2010-07-29 出版日期:2010-12-25 发布日期:2010-12-25
  • 通讯作者: 温惠英
  • 作者简介:温惠英(1965-), 女, 江西省于都人, 教授, 博士.

Study of Decision Tree Structure of Traffic Mode Choice Based on the Goodness of Fit Measure

WEN Hui-ying; ZENG Wen-chuang   

  1. Shool of Civil Engineering and Transportation, South China University of Technology, Guangzhou 510640, China
  • Received:2010-03-31 Revised:2010-07-29 Online:2010-12-25 Published:2010-12-25
  • Contact: WEN Hui-ying

摘要: Nested-Logit(NL)模型是交通方式划分的最重要的模型之一,在分析公共交通和个人交通划分方法的理论依据的基础上,根据人们在选择交通方式过程中首要考虑因素的不同,构造了不同于公共交通和个人交通的虚拟交通方式——高费用和低费用虚拟交通方式,并分析其内相关性. 在两个交通方式选择决策树结构都存在相关性和合理性的情况下,提出基于拟合优度度量比较两者的优劣,从而选定交通方式选择决策树结构. 根据实例旅客出行调查数据,利用SAS软件实现两个交通方式选择决策树结构的拟合度度量分析. 结果表明:基于拟合优度度量选定的高费用和低费用虚拟交通方式比公共交通和个人交通的虚拟交通方式更接近实际调查值,此方法具有一定的实用性.

关键词: 城市交通, 决策树结构, NL模型, 拟合优度度量, 交通方式之间的相关性

Abstract: Nested-Logit model (NL model) is one of the most important models in traffic modal split, based on theoretical basis of the method of traffic modal split for public traffic and private traffic, according to the different factors of people’s first considering in the middle part, the paper constructed virtual traffic modes different from public traffic and private traffic—high-cost and low-cost virtual traffic modes, and analyzed the correlation of alternatives in virtual traffic. the paper also proposed method of choosing the better decision tree structure of traffic mode choice based on goodness of fit measures. According to passenger travel survey data in the example, using SAS software, the paper comparatively analyzed the goodness of fit measure, and the results showed that: the choice probabilities from high-cost and low-cost virtual traffic modes based on the goodness of fit measure can better closer to the survey value. This method is useful in application.

Key words: urban traffic, decision tree structure;NL mode;goodness of fit measure;correlation among traffic models

中图分类号: