Free interchange for better transit? Assessing the multi-dimensional impacts on metro to bus interchange behavior − insights from an explainable machine learning method

This study investigates the impact of a newly implemented public transport interchange discount policy in Suzhou, China, focusing on its effects on metro-to-bus interchange behaviors across various spatial and temporal dimensions. Utilizing two distinct datasets spanning periods before and after the policy's implementation, a comprehensive spatial–temporal analysis was conducted, covering weekdays, weekends, and holidays. A novel, real-time, distance-weighted methodology was developed to more accurately identify metro-to-bus interchange catchments, thereby refining the modeling scope. The study examines the interplay between land use, socio-demographic factors, and bus-related attributes—including a newly proposed operation-opportunity combined bus accessibility metric—using an explainable machine learning approach. Results indicate that the interchange discount policy has had an overall positive, though varied, impact on interchange behaviors, with the most pronounced effects observed during weekdays in central urban areas and at metro line bends. Specifically, 76.1 % of metro stations saw an increase in metro-to-bus interchange ratios on weekdays following the policy's implementation, with increases observed at 66.4 % and 67.3 % of stations during weekends and holidays, respectively. Overall, the interchange ratio increased by 12.49 %, with a 17.45 % increase on weekdays. The analysis also reveals that factors such as bus accessibility, bus-to-bus interchange, and population density exhibit different effects depending on the time of week, with non-linear patterns emerging. The policy's introduction shifted the impact thresholds for certain factors, initially triggering competition between bus and metro services but eventually leading to a synergistic rise in metro-to-bus transfers as bus-to-bus interchange ratios increased. Additionally, the policy altered the significance of population density, enhancing the attractiveness of multimodal interchange for users who previously favored other modes of transport.