Abstract:

Accurately characterizing the relationship between the built environment and urban rail transit ridership is an important prerequisite for understanding passenger demand. In response to the challenges of incomplete and high-dimensional sparse data in inter-station OD (origin-destination) studies, this paper proposed a research method of mapping relationship between built environment and ridership at the zone level. Firstly, a two-level station clustering method was developed by replacing individual stations with clusters (“cluster-over-point”) based on natural geographical characteristics and passenger flow destination features. Inter-cluster similarity was calculated to a-ddress the issue of data sparsity. Secondly, a built environment indicator system and corresponding description method were constructed from two dimensions: the attraction capacity of origin or destination clusters and OD acce-ssibility characteristics. Thirdly, a methodology based on the Gradient Boosting Decision Tree (GBDT) model was introduced to characterize the relationship between built environment features and passenger flow, delving into the influence intensity and threshold values of individual factors on passenger flow. Finally, the proposed method was validated using data from the Beijing Subway. Therefore, in the process of urban rail transit planning, priority should be given to optimizing network topology and improving transportation accessibility, followed by a deeper consideration of the impact of regional economic activities. The results show that the mapping relationship between built environment and passenger flow at zone-to-zone level has spatial and temporal heterogeneity, nonlinear characteristics and threshold effects. The zoning-based research perspective effectively addresses issues of data sparsity. OD impedance emerges as the primary feature influencing passenger flow, accounting for up to 38.40% of the explanatory power, while demographic and economic characteristics serve as secondary factors, exhibiting significant threshold effects. Therefore, in the process of urban rail transit planning, priority should be given to optimizing network topology and improving transportation accessibility, fo-llowed by a deeper consideration of the impact of regional economic activities. The research findings provide quantitative analytical tools for rail transit planners, assisting them in identifying the effective ranges of built environment indicators and adjusting spatial configurations. These insights offer valuable references for enhancing the operational efficiency of urban rail systems.

Key words: urban rail traffic, built environment, gradient boosting decision tree model, passenger flow at zone-to-zone level, nonlinear relationship