Uncovering spatial patterns of environmental influence on the paces of active leisure travel

Abstract

Exploring the environmental influence patterns on active leisure travel (ALT) is crucial for comprehending how spatial contexts shape travel behavior. Although previous studies have linked environmental factors to ALT participation, their influence on ALT paces across different modes remains underexplored. This study collects 10,649 crowdsourced GPS trajectory datasets to depict the distribution of walking, jogging, and cycling paces in Shenzhen. By employing an explainable machine learning framework integrating XGBoost and SHAP, we reveal the nonlinear effects of both natural and built environments on ALT paces. Additionally, a k-means cluster analysis is applied to identify impact patterns among geographical units. Our results indicate that environmental features have different impact contributions on the paces of different travel modes, all demonstrating nonlinear relationships with thresholds. Moreover, influence patterns show significant variation across local regions, reflecting diverse environmental preferences for travel paces. These findings offer valuable insights for urban planners to implement context-specific interventions that foster ALT-friendly urban environments.