Examining the effects of station-level factors on metro ridership using multiscale geographically weighted regression

Abstract

Metro systems provide a mass and sustainable mobility option for urban populations. Drawing on the ridership data in 2019 released by the Shanghai Municipal Transportation Commission, we examine the spatially varying effects of station-level factors measured with different metro-station catchments (MSCs) on the daily and hourly ridership through multiscale geographically weighted regression (MGWR) models. The results indicate that independent variables measured with relatively small and non-overlapping MSCs can better explain the spatial variations in metro ridership. Regarding the important determinants, land use intensity (LUI) demonstrates positive effects diminishing from the core to the peripheral areas, transfer lines (TL) also exhibits positive effects that decrease from southwest to northeast. The effects of road density (RD) and building density (BD) are negative in central urban areas but positive in suburbs, which is the opposite of bus lines (BL). Additionally, the explanatory variables have significantly different ranges of influence, and the range of influence for a single variable also varies across models that use ridership at different times (e.g., weekdays vs. non-weekdays, the morning/evening peak vs. non-peak hours) as dependent variables. This study expands the knowledge of the spatial variations in the relationship between metro ridership and the geographical determinants. The findings can provide empirical evidence and implications for urban planners in formulating context-specific policies to improve the usage of metro systems.