Exploring nonlinear associations between environmental and social media variables and active leisure travel

In the context of healthy urban development, active leisure travel modes such as cycling, jogging, and walking promote physical health and reduce health risks. Current research has predominantly focused on the associations between environmental variables and active leisure travel, while limited attention has been given to public perceptions and online interaction characteristics captured through social media. This study expands the existing environmental variable framework by incorporating social media variables. It aims to investigate the nonlinear associations between multidimensional variables and different types of active leisure travel. This study utilized 5,451 active leisure travel trajectory records to develop cycling, jogging, and walking indices, which were used as dependent variables. The Chinese-RoBERTa-wwm-ext model was utilized to analyze social media data, producing emotional scores and interaction heat variables. Environmental and social media variables were integrated, and the XGBoost model combined with the SHAP interpretation approach was employed to examine their nonlinear associations with active leisure travel. The three active leisure travel modes exhibit notable spatiotemporal heterogeneity, characterized by considerable variations in key explanatory variables and contribution patterns. For cycling, interaction heat, average temperature, and NDVI were the variables with relatively higher contributions; for jogging, population density and spatial accessibility showed greater importance; and for walking, average temperature, interaction heat, and emotional scores exhibited relatively higher contributions. All variables demonstrate distinct nonlinear associations and threshold patterns with active leisure travel, with their significance and explanatory capacity varying across different travel modes. Furthermore, social media characteristics make significant contributions. The results enhance our comprehension of active leisure travel and offer a novel viewpoint for research on urban leisure travel.