Variance inflation factor test table.

The built environment is an important determinant of travel demand and mode choice. Studying the relationship between the built environment and transportation usage can support and assist traffic policy interventions. Previous studies often assumed that this relationship is linear; however, the impact of the built environment on non-motorized travel efficiency may be more complex than the typically modeled linear relationships. This paper focuses on the core area of Chengguan District in Lanzhou City, utilizing multi-source big data including POI, OpenStreetMap, street view images, and built environment data. Using ArcGIS spatial analysis tools combined with the Extreme Gradient Boosting (XGBoost) model, we analyze the non-linear influence mechanisms and threshold effects of the built environment on non-motorized travel efficiency and establish a ranking of the relative importance of all built environment factors. The results indicate that factors such as the branch road/street, land-use mix, land-use density, neighborhood entrance/exit density, bus station density, and dead-end-roads density are key influences on non-motorized travel efficiency. Additionally, based on the non-linear thresholds presented in the partial dependence plots for built environment factors, this paper proposes optimization strategies for small-scale road network patterns, mixed land use, and bus-friendly environments, providing effective threshold ranges and decision-making references for urban planning and traffic management.