Passenger Travel Behavior in Public Transport Corridor After the Operation of Urban Rail Transit: A Random Forest Algorithm Approach

Yanqing Liao,Huizhen Zhang,Xiaofei Li,Yueer Gao

doi:10.1109/access.2020.3038831

Abstract

A competitive relationship has been generated between urban rail transit and bus transit since the operation of the former. Despite different roles in providing services in public transport corridor, they affect each other in actual situations. In terms of urban transportation planning and policy formulation, it is necessary to explore and master the rules of passengers' travel mode choice from different means. In order to study their choices between the rail transit and the bus transit after the operation of the former, taking Xiamen, China as an example, this article analyzed the overall travel features of passenger flow before and after the operation of rail transit by using the public transit IC card data from two consecutive weeks in November 2017 and November 2018. Some features of travel distance, travel time, travel cost, whether to travel in peak hours, the number of collinear stations between bus transit and rail transit or that of rail transit stations are sorted out. With random forest algorithm, a model is set up for the travel mode choice of passengers after urban rail transit is put into use to find out the impact of different travel features. The result shows that travel cost is the most crucial factor that affects passengers' decisions, followed by the number of collinear stations between bus transit and rail transit or that of rail transit stations, travel time and travel distance. Whether to travel in peak hours have less impact on their choices. This study is constructive for cities in the stage of facing competition between newly opened rail transit and bus transit and support transportation decision-making.

Highlights

Urban rail transit is built up in an increasing number of cities with the aim of alleviating the deteriorating traffic congestion with a public transport corridor operating mainly with the rail transit and supplemented with the bus transit [1][2]
Exploring the travel rules of passengers with their travel behaviors can lead to targeted solutions for small passenger flow in some rail transit routes caused by the competition of different travel mode in public transport corridor, which can improve its transit efficiency and promote urban development
In order to better compare the impact of various features on travel mode choice, the importance of different features in the data samples is evaluated during the establishment of the random forest algorithm model to analyze their impact on the sample classification or in other words, on passengers’ choices

Summary

INTRODUCTION

Urban rail transit is built up in an increasing number of cities with the aim of alleviating the deteriorating traffic congestion with a public transport corridor operating mainly with the rail transit and supplemented with the bus transit [1][2]. Background, this paper studies passengers’ travel mode choice in the public transport corridor between the bus transit and the rail transit after the operation of the latter in Xiamen in 2018. In addition to the travel time, travel distance, travel cost, and whether to travel in peak hours mentioned earlier in this paper, the number of collinear stations between bus transit and rail transit or that of rail transit stations is used to establish a model with random forest algorithm for passengers’ travel modes choice after the operation of the rail transit to analyze the importance of various factors that affect passengers’ travel mode choice.

Data Sources

B Data Processing

C Experiment Methods

Data Analysis and Research Results

Discussions

Findings

Conclusions