Journal of Transportation Systems Engineering and Information Technology

Journal of Transportation Systems Engineering and Information Technology ›› 2020, Vol. 20 ›› Issue (4): 136-142.

Previous Articles     Next Articles

Identification Method of Passenger's Dependence Level of Public Transportation Based on Correlation Analysis

HU Song1 , WENG Jian-cheng1 , ZHOU Wei2 , LIN Peng-fei1 , KONG Ning1   

  1. 1. Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing 100124, China; 2. Ministry of Transport of People's Republic of China, Beijing 100736, China
  • Received:2020-03-23 Revised:2020-06-09 Online:2020-08-25 Published:2020-08-25

基于关联分析的乘客公共交通依赖度识别方法

胡 松1 ，翁剑成*1 ，周 伟2 ，林鹏飞1 ，孔 宁1   

  1. 1.北京工业大学 北京市交通工程重点实验室，北京 100124；2.中华人民共和国交通运输部，北京 100736
  • 作者简介:胡松(1992-)，男，河北邢台人，博士生.
  • 基金资助:

    国家自然科学基金重大项目/ Major Program of the National Natural Science Foundation of China(U1811463)；国家自然科学基金/National Natural Science Foundation of China(51578028)；北京市“科技新星”计划“/ Beijing Nova”Program by Beijing Municipal Science and Technology Commission(Z171100001117100).

PDF

Abstract:

There are significant variations in the dependence degree on public transportation (PT) for passengers in their long-term travel behavior. Accurate identification of the PT dependence level of passengers is conducive to promote the attractiveness of public transportation. The PT travel chain containing individual attributes is obtained from multi- source heterogeneous data, then 8 indicators are proposed to measure the travelers' PT dependence from the aspects of individual travel behaviors and attribute information. Thus, a two- step clustering model is constructed to identify the PT dependence level of passengers. The results show that the respondents are clustered into 4 groups according to the travel dependence levels of public transport. The respondents with the high PT dependence level are limited by their income and vehicle ownership, and there is a trend of transfer to private cars. In addition, travel habit behavior has a greater impact on the individual PT dependence level compared with the individual attributes of passengers. Finally, the evaluation indicators of AHR and ACR are used to further evaluate the impacts of individual attributes on recognition results. The results show that the individual attribute indicators have coupling effects on the model results, and the indicators missing quantity and model error own the nonlinear relationship. This study is conducive to better understanding the passengers' travel rules and demands, and provides support for the improvement of public transport services.

Key words: urban traffic, identification of travel dependence, two-step clustering model, travel behavior of public transport, data correlation

摘要：

不同乘客在出行过程中对公共交通的依赖程度具有显著差异，精准识别乘客公共交通依赖度，有助于针对性地引导出行者向公共交通方式转移.本文基于多源数据的关联，获取包含个体属性的公共交通出行链，从出行行为和个体属性两方面提出8个依赖度指标，构建二阶聚类模型，识别乘客公共交通依赖度.结果表明：样本按依赖度高低被划分为4类群组；部分高、较高依赖度乘客在出行决策时受限于收入和车辆拥有量，并有向私家车出行转移的趋势；乘客出行习惯行为较个体属性对公共交通依赖度的影响更大.利用平均命中率(AHR)和平均覆盖率(ACR)指标评估个体属性对识别结果的影响，得出结论，个体属性指标间存在耦合关系，且指标缺失量与模型误差具有非线性关系.研究有助于理解公共交通乘客的需求和选择倾向性，为精准改善公共交通服务提供支撑.

关键词: 城市交通, 依赖度识别, 二阶聚类模型, 公交出行行为, 数据关联

CLC Number: