Micro-level walkability and station-level transit use: Integrating perceptual and spatial measures in Guiyang, China

Despite growing investments in Urban Rail Transit (URT), many station areas remain poorly designed for pedestrian access, limiting the effectiveness of transit-oriented development (TOD). While previous studies have emphasized meso‑scale built environment factors, limited research has quantitatively evaluated how micro-level walkability features affect transit ridership. To address this gap, this study proposes a composite evaluation model to assess the walkability of station neighborhoods using street-view imagery (SVIs) and computer vision tools. The model incorporates three key dimensions, mobility, safety and security, and comfort and pleasantness, identified through a survey of metro users in Guiyang, China. The relative importance of these dimensions was derived using Structural Equation Modeling (SEM). Streetscape features extracted from SVIs were aggregated to compute walkability scores for 800-meter station buffer areas. The association between walkability and average weekday metro ridership was examined using Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR), followed by K-means clustering to categorize station types. Results confirm that both microand meso‑level features significantly influence ridership. The proposed method provides a scalable tool for diagnosing walkability gaps and informing differentiated planning strategies to improve access, particularly in medium-sized cities undergoing rapid URT expansion. © 2025 Elsevier Ltd.