Nonlinear Analysis of Commuting Performance and Its Influencing Factors in Mountainous Cities: Implications for Sustainable Transportation

Improving commuting performance is critical for high-quality urban transportation development and residents’ well-being. Mountainous cities face more stringent spatial constraints on commuting due to terrain barriers. However, most existing studies on commuting performance ignore the terrain factors, with few evaluation frameworks tailored to mountainous contexts and insufficient research on the nonlinear mechanisms of driving factors. Taking central Chongqing, a typical mountain-valley metropolis, as the study area, this paper constructs a four-dimensional commuting performance evaluation system (efficiency, equity, terrain adaptability, organization) covering 155 commuting zones from a dual jobs–housing perspective. A random forest model combined with SHAP interpretable analysis is employed to identify core drivers, nonlinear effect patterns, and factor interactions. The results reveal significant zonal disparities shaped by mountain terrain: high-performance clusters in flat central valley areas, while mountainous belts remain at low levels. Temporally, overall commuting performance declined from 2017 to a trough in 2021, followed by a slight rebound by 2023. Employment density, topographic relief, residential density, and geographic location collectively account for 75.4% of the total relative feature importance in the model, whereas micro built environment factors exert weak independent effects on commuting performance. All core variables exhibit threshold-based nonlinearity, and jobs–housing co-agglomeration generates the strongest positive synergy. This study clarifies the driving mechanisms underlying commuting performance in mountainous cities and provides empirical references for urban spatial planning and commuting governance, which further promotes sustainable, high-quality urban commuting in such areas.