Environmental factors and social media metrics exhibit distinct nonlinear associations and threshold patterns with active leisure travel, according to a study published in the International Journal of Health Geographics. Researchers Xueying Huang, Jun Zhang, and Jia Xu found that the explanatory capacity and relative importance of multidimensional variables vary considerably among different physical travel modes, demonstrating notable spatiotemporal heterogeneity across urban environments.

    The authors framed their investigation around healthy urban development, observing that active leisure travel modes such as cycling, jogging, and walking promote physical health and reduce health risks. Existing literature has predominantly concentrated on relationships between environmental variables and active leisure travel, while public perceptions and online interaction characteristics captured through social media have received limited scholarly attention. To address this gap, the study expanded the traditional environmental variable framework by incorporating social media variables.

    To capture movement patterns, the research team analyzed 5,451 active leisure travel trajectory records. These trajectory records were utilized to construct specific cycling, jogging, and walking indices, which served as the primary dependent variables in the empirical models. By distinguishing between these three travel forms, the authors examined how specific environmental and digital factors nonlinearly correlate with different types of active leisure engagement.

    To incorporate the social media dimension, the researchers processed social media data using the Chinese-RoBERTa-wwm-ext natural language processing model. This computational analysis yielded two central social media variables: emotional scores and interaction heat. The authors subsequently integrated these digital interaction metrics with physical environmental variables, establishing a multidimensional dataset designed to reflect both physical surroundings and public online engagement.

    For their analytical framework, the authors integrated the environmental and social media datasets using an XGBoost machine-learning model combined with the SHAP (SHapley Additive exPlanations) interpretation method. This modeling approach was applied to examine nonlinear associations and threshold patterns across the active travel types. SHAP values allowed the researchers to interpret variable contribution patterns and assess associations with leisure travel outcomes.

    The findings revealed considerable variations in key explanatory variables and their relative contribution patterns across the three travel modes. For cycling, interaction heat from social media, average temperature, and the normalized difference vegetation index (NDVI) represented the variables with relatively higher contributions. In contrast, jogging behaviors showed higher importance for population density and spatial accessibility in the models.

    For walking, average temperature, interaction heat, and emotional scores exhibited relatively higher contributions. Across all three modes, variables demonstrated distinct nonlinear relationships and clear threshold patterns rather than linear effects. The authors concluded that social media characteristics make significant contributions to explaining active leisure travel, offering a novel viewpoint for future research on urban leisure travel dynamics.

    Subject of Research: Nonlinear associations of environmental and social media variables with active leisure travel

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

    Article References: Huang, X., Zhang, J., & Xu, J. (2026). Exploring nonlinear associations between environmental and social media variables and active leisure travel. International Journal of Health Geographics. https://doi.org/10.1186/s12942-026-00496-0

    Image Credits: AI Generated

    DOI: 10.1186/s12942-026-00496-0

    Keywords: active leisure travel, cycling, jogging, walking, environmental variables, social media variables, interaction heat, emotional scores, Chinese-RoBERTa-wwm-ext, XGBoost, SHAP interpretation, average temperature, NDVI, population density, spatial accessibility, spatiotemporal heterogeneity, nonlinear associations, threshold patterns

    Share.

    Comments are closed.