Built Environment Inequality and the Spatial Structure of Digital Mobility Networks Evidence from Shared Bicycle Systems in Shenzhen

Abstract

Shared bicycle systems are increasingly central to sustainable urban mobility, yet they often reflect and amplify spatial inequalities in the built environment. Existing studies focus on static correlations between urban features and mobility, relying on linear models and uniform spatial assumptions, leaving unclear how local inequalities shape the emergent structure of digital mobility networks. This study investigates how built environment inequality shapes the spatial structure of digital mobility networks, using dockless bike-sharing in Shenzhen as an empirical case. To capture complex nonlinear and spatially heterogeneous effects, we employ a hybrid analytical strategy integrating XGBoost-SHAP for feature importance and nonlinear interpretation, and Geographically Weighted Gaussian Process Regression for continuous spatial heterogeneity analysis. Our findings reveal that built environment inequality nonlinearly and spatially heterogeneously molds digital mobility networks. Inequalities in transport and service facilities emerge as dominant drivers, producing monotonic, threshold, and saturation effects that vary across network dimensions. Core urban areas experience amplified network intensity, connectivity, and destination diversity, while peripheral zones show attenuated or suppressive effects, reflecting a pronounced core-periphery gradient. These findings provide empirical evidence for designing inclusive, network-oriented urban mobility policies that account for multidimensional and spatially contingent inequalities.