This study takes the Guiyang Loop Line Railway as a case study and, based on resident commuting survey data, applies an XGBoost model combined with Individual Conditional Expectation (ICE) analysis to examine the threshold effects and non-linear spatial characteristics of the built environment on residents' willingness to commute via the Loop Line Railway. The results indicate that: (1) population density, distance to the nearest bus stop, road network density, and distance to the city center are key built environment factors influencing commuting willingness. Among them, population density and road network density exhibit overall positive effects, whereas distance to the nearest bus stop and distance to the city center show negative effects; (2) the spatial distributions of the threshold effect magnitude (Δy) and the non-linear effect intensity (m) display a heterogeneous pattern broadly characterized by a "city center-suburban-peripheral" gradient, suggesting pronounced spatial non-stationarity in the relationship between the built environment and commuting willingness; (3) compared with the central urban area, residents living around suburban and peripheral stations are more sensitive to changes in bus feeder conditions, road network connectivity, and commuting distance, exhibiting stronger threshold effects and non-linear responses. These findings suggest that enhancing residents' willingness to commute via the Loop Line Railway requires context-sensitive strategies that account for the built environment characteristics of different areas, with targeted improvements in bus-rail integration and road network conditions, rather than adopting uniform development and supporting policies across all stations.
Chen et al. (2026) studied this question.