Investigating intercity commuting using distance decay in a two-stage regression framework: A case study of Shenzhen Metropolitan Area in China

Metropolitan areas are the most concentrated spaces of population and economic activity. Intercity commuting is not only a core driver of metropolitan operations but also an important indicator of metropolitan integration. However, existing studies lack a systematic framework to characterize intercity commuting and compare it with intracity commuting. This study uses mobile phone signaling data from the Shenzhen Metropolitan Area and proposes a two-stage regression framework. In the first stage, a flow-based geographically weighted regression model is used to estimate distance decay coefficients; in the second stage, random forest combined with SHAP analysis is applied to examine the influence of the built environment. The results show that intercity commuting decays faster than intracity commuting and exhibits spatial heterogeneity. Along city boundaries, short-distance intercity commuting shows quick decay due to limited transport supply and administrative barriers; along the perpendicular direction, medium- and long-distance intercity commuting is influenced by cross-border coordination, high-level transport infrastructure, and housing price differences, which reduce decay, but poor connectivity or high combined costs of housing and transport can still increase it.