Exploring the spatiotemporally varying effects on urban rail transit station ridership: A Novel Methodological Framework

Previous studies using multi-source big data primarily interpreted spatiotemporal variations in urban rail transit station ridership and its influencing factors as spatiotemporal heterogeneity. However, these studies often lack precise metrics to quantify this spatiotemporal heterogeneity. To address this gap, our study introduces an integrated framework comprising a novel spatiotemporal weighted regression (STWR) model, Theil index (TI), and time series entropy (TSE). An empirical study is conducted in Hangzhou, China, using three-week dataset encompassing smart card transactions and station-level point-of-interest (POI) data. Comparative results indicate that the STWR model exhibits superior performance in comparison to the classic geographically temporally weighted regression (GTWR) model. Building on this foundation, we investigate the spatiotemporal interactions between significant influencing factors and station ridership from both macro and micro perspectives, incorporating urban land use, transport connection accessibility, station classification, and ridership patterns. Furthermore, the TI results reveal that the spatial variation in influencing factors’ effects on station ridership is particularly heterogeneous in urban central area and certain suburban areas, which is driven by the uneven land development, infrastructure configuration, and rail transit network layout in these areas. The magnitude of temporal heterogeneity quantified by TSE is closely associated with the number and shape of peaks and troughs in the time series of estimated coefficients for the effects. Our findings can provide valuable insights into optimal allocation of urban rail transit resources and support the coordinated development of urban public transportation and land use.