Jihao Deng , Yiting Yang , Tianhao Li , Lei Gao , Chris Bachmann , Xiaohong Chen
{"title":"Unveiling Route Choice Preferences and Classifying Travelers Based on Distinct Travel Patterns Using Trajectory Data","authors":"Jihao Deng , Yiting Yang , Tianhao Li , Lei Gao , Chris Bachmann , Xiaohong Chen","doi":"10.1016/j.cstp.2025.101399","DOIUrl":null,"url":null,"abstract":"<div><div>Effective travel demand management strategies hinge on a better understanding of route choice behavior. This complexity is further intensified by the diversity of activity patterns and the intricate structure of urban networks, posing significant challenges to conducting and interpreting large-scale, detailed analyses of drivers’ travel preferences. The rapid adoption of electric vehicles (EVs) equipped with positioning systems enables the collection of large-scale individual travel data and the observation of repetitive travel patterns, offering new opportunities to analyze travel preferences. Utilizing anonymized trajectory data from 4,551 family-operated EVs in Shanghai, this study delves into travelers’ route choice preferences and identifies distinct traveler types. We introduce the concept of an entropy-weighted propensity score to capture the heterogeneity in travelers’ preferences for various route characteristics and examine these preferences across different travel scenarios – including trip distance, trip types, periods of departure time, origin and destination locations, and land use mix. Furthermore, we propose a sample-weighted K-means clustering method to categorize travelers into three distinct types: those exhibiting high travel variability with low rationality, low travel variability with medium rationality, and high travel variability with high rationality. Our findings deepen the current understanding of route choice behaviors and could provide empirical support for enhancing the precision and effectiveness of route choice models. When examining route choice behavior, it is essential to account for conventional fuel vehicles and to critically assess the distinctions between them and EVs. Moreover, they offer practical insights for alleviating traffic congestion through personalized active traffic demand management strategies.</div></div>","PeriodicalId":46989,"journal":{"name":"Case Studies on Transport Policy","volume":"19 ","pages":"Article 101399"},"PeriodicalIF":2.4000,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Case Studies on Transport Policy","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2213624X25000367","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"TRANSPORTATION","Score":null,"Total":0}
引用次数: 0
Abstract
Effective travel demand management strategies hinge on a better understanding of route choice behavior. This complexity is further intensified by the diversity of activity patterns and the intricate structure of urban networks, posing significant challenges to conducting and interpreting large-scale, detailed analyses of drivers’ travel preferences. The rapid adoption of electric vehicles (EVs) equipped with positioning systems enables the collection of large-scale individual travel data and the observation of repetitive travel patterns, offering new opportunities to analyze travel preferences. Utilizing anonymized trajectory data from 4,551 family-operated EVs in Shanghai, this study delves into travelers’ route choice preferences and identifies distinct traveler types. We introduce the concept of an entropy-weighted propensity score to capture the heterogeneity in travelers’ preferences for various route characteristics and examine these preferences across different travel scenarios – including trip distance, trip types, periods of departure time, origin and destination locations, and land use mix. Furthermore, we propose a sample-weighted K-means clustering method to categorize travelers into three distinct types: those exhibiting high travel variability with low rationality, low travel variability with medium rationality, and high travel variability with high rationality. Our findings deepen the current understanding of route choice behaviors and could provide empirical support for enhancing the precision and effectiveness of route choice models. When examining route choice behavior, it is essential to account for conventional fuel vehicles and to critically assess the distinctions between them and EVs. Moreover, they offer practical insights for alleviating traffic congestion through personalized active traffic demand management strategies.