Age disparities and socioeconomic factors for commuting distance in Beijing by explainable machine learning

The increasing commuting issues faced by residents in China's megacities have led to a growing body of literature on commuting equality. However, longitudinal evidence on the heterogeneous and nonlinear associations between socioeconomics and commuting distances across different age groups remains unknown. This study employs a large-scale dataset of location-based data from mobile devices to identify age disparities in commuting patterns, home-work balance, and commuting distance. We take the commuting patterns in Beijing as a case study. Employing the eXtreme Gradient Boosting (XGBoost) machine learning model and the Shapley Additive exPlanations (SHAP) method, we examined and explained the nonlinear interactive effects of individual and socioeconomic characteristics on commuting distance. The results revealed significant age disparities in the work-home balance within Beijing, with young individuals tending to have longer intra-city commuting distances than the old. This study highlights the impacts of individual and socioeconomic attributes on commuting disparities across age groups. Housing prices emerged as the most significant factor explaining commuting distance, followed by the importance of achieving a suitable home-work balance for young people. The spatial contradiction between housing and employment opportunities has played a crucial role in shaping commuting patterns. These insights contribute to urban planning efforts aimed at achieving social equity in commuting and enhancing the overall quality of life in cities.