Non-linear impact of built environment on origin–destination passenger flow in Tianjin’s urban rail transit

It is essential to understand the impact of the built environment on origin–destination (OD) ridership, as this understanding is crucial for the optimisation of urban rail systems and the advancement of transit-oriented development (TOD) strategies. This study utilises an eXtreme gradient boosting (XGBoost) model to analyse non-linear relationships across 16 built environment factors, including population density, land use and transportation access, influencing OD flows on weekdays and weekends. The analysis, employing the Tianjin Metro as a case study, yielded three key findings. The influencing factors at origin and destination stations exhibit temporal variations. During weekday morning peak hours, the explanatory variables at origin stations hold greater importance than those at destination stations, while the opposite is true for weekday evening peaks and weekends. It is demonstrated that land use and transportation connectivity are among the most significant factors, with an importance score of 3.50% at both origin and destination stations, thereby underscoring their substantial impact on ridership attraction. Non-linear interactions reveal threshold effects, emphasising the complex relationship between built environment characteristics and OD flows. The insights derived from this analysis offer practical guidance for urban planners, informing the design of efficient rail systems and the integration of TOD strategies.