How do built environment characteristics influence metro-bus transfer patterns across metro station types in Shanghai?

Abstract

Unravelling the complex relationship between metro-bus transfer behavior and the built environment is crucial for the construction of a sustainable urban public transportation system. The current research prominently emphasizes modeling station-level metro-bus transfer ridership in relation to the built environment that surrounds with transit stations, few has specially focused on exploring and comparing this relationship among various transit station types. Based on the case study of Shanghai central city, this research clustered metro stations according to the time-series similarity of metro-bus transfer ridership pattern by combining Derivative Dynamic Time Warping and K-medoids. Then, for each metro station group, the spatiotemporal heterogeneity and nonlinearity of built environment effects on transfer ridership pattern were examined simultaneously by applying an adapted GTWR-RF method that integrates Geographically and Temporally Weighted Regression (GTWR) and Random Forest (RF). Our empirical analysis confirmed the importance of key built environment determinants and their associations with transfer ridership vary significantly among different metro station types. Furthermore, this research highlighted the proposed GTWR-RF model, which considers both spatiotemporal heterogeneity and nonlinearity effects of the built environment on the transfer ridership, can significantly improve the prediction ability. These findings provide a comprehensive perspective for policymakers, enabling them to formulate transportation policies with consideration of station type specification and to bolster the overall public transportation usage in cities.