Enhancing public transport competitiveness in metropolitan areas: Empirical evidence and policy implications from Beijing's commuting

Rapid metropolitan integration has intensified commuting within metropolitan areas, yet persistent car dependence undermines low-carbon mobility goals. This study investigates the determinants of travel mode choice in the Beijing metropolitan area, with a particular focus on the underexplored role of service quality. Drawing on China's first large-scale metropolitan travel survey (n = 3069) and employing an interpretable Light Gradient Boosting Machine (LightGBM) model with SHapley Additive exPlanations (SHAP) analysis, we examine how socioeconomic attributes, built environment, and service quality influence commuting choices. Result stability is further assessed via bootstrap resampling. The results show that service quality is pivotal for public transport uptake, with pronounced nonlinear and threshold effects associated with bus detour ratio, transfer distance, walking distance to stops, and fares. Car ownership, especially possession of a Beijing license plate, remains the strongest single determinant of commuting mode choice. The nonlinear analysis identifies critical breakpoints: for example, walking distance to bus stops exceeding 700 m or detour ratios above 1.4 markedly reduce the probability of choosing public transport. Interaction effects indicate that improving stop accessibility and promoting amenity clustering around workplaces can strengthen the competitiveness of public transport in metropolitan commuting. These findings offer actionable policy implications. Enhancing public transport competitiveness necessitates reducing detours and transfer distances, improving access to stops, and maintaining affordable fare levels, potentially through targeted subsidies. At the spatial level, implementing Transit-Oriented Development (TOD) strategies can reinforce the synergies between concentrated land use and efficient transit services. Overall, the study provides empirical evidence to support metropolitan public transport policy design and offers methodological insights that can inform the calibration of discrete choice models.