Understanding the impact of built environment on metro ridership using open source in Shanghai

A growing body of research using the direct demand model has explored the impact of the built environment on transit ridership. However, empirical studies identified various significant factors in different cities with different datasets. This study adopts points-of-interest (POIs) data to identify the physical environmental factors affecting metro ridership in Shanghai. Independent variables in terms of the rail transit system, external connectivity, intermodal connection, and land use factors within 286 metro stations' catchment areas were selected. Principal component analysis (PCA) was used to group POIs into 6 components for dimensionality reduction. The results from ordinary least squares (OLS) regression analysis emphasize the dominating role of commercial land use and rail transit system factors, together with bus stops, tourist spots and healthcare factors, positively impact both weekday and weekend metro ridership; however, the effect of job-related land use is significant only on weekdays. Distinctively, the variable of intersection density is not positively associated with ridership as expected, revealing that street network measurements may not explain walking to rail transit in the citywide Shanghai context, so we suggest a new requirement: a multilevel-based walkability index in dense cities. The latter finding also implied that residences in central locations are less reliable than those in suburban locations. Finally, we conclude with strategies to encourage balanced trip demands other than simply increasing ridership, which has potential implications on urban planning and transit-oriented development (TOD) in China.