Nonlinear relationship between built environment and commute mode choice: A comparison of commuters within and beyond the long commute threshold

• Published in: Journal of transport geography Vol. 129; p. 104430
• Main Authors: Yu, Ming, Tong, Zhaomin, Yang, Jiaming, Lu, Yanchi, Liu, Yang, Liu, Yanfang, Liu, Yaolin
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.12.2025
• Subjects: ALE algorithm; Built environment; Commute mode choice; LightGBM model; Nonlinear effect; SHAP algorithm; SHAP algorithm; Nonlinear effect; LightGBM model; Built environment; Commute mode choice; ALE algorithm
• ISSN: 0966-6923
• DOI: 10.1016/j.jtrangeo.2025.104430

This study uses machine learning to explore the nonlinear relationship between the built environment and the commute mode choice of commuters within (Commuter I) and beyond (Commuter II) a long commute threshold. Utilizing 739,457 location-based service data from Wuhan, a method that determines the long commute threshold through the marginal contribution of green commute time to car commuting is introduced. Commuters are then divided into Commuter I and Commuter II based on the long commute threshold. Subsequently, the light gradient boosting machine and accumulated local effect methods are employed to identify the nonlinear relationship between commute mode choice and the built environment for the two groups. Results show that in Wuhan, the long commute thresholds are 35 min for car commuters and 45 min for green commuters. For Commuter II, most built environment indicators reflect socioeconomic characteristics, suggesting that the built environment has a more limited moderating effect on their mode choice compared to Commuter I. For Commuter I, their probability of car commuting may be reduced by enhancing residential land-use diversity, increasing employment opportunities, and implementing a “narrow streets, dense network” infrastructure. Therefore, the proportion of commuters with green commutes under 35 min should be increased in the future to maximize the role of the built environment in regulating car commuting. These findings can provide a detailed reference for the green transformation and sustainable development of urban transportation. [Display omitted] •Commute mode choice is well identified by actual and green commute time.•The long commute threshold (car commuter: 35 min, green commuter: 45 min) in Wuhan is determined.•Built environment affects commuters differently within/beyond long commute threshold.•Threshold analysis of built environment can provide reference for sustainable urban development.