Measuring the impact of built environment factors on station-level contributions to link-level crowding using a novel crowding contribution index

Abstract

Abstract Metro crowding undermines passenger comfort, operational efficiency and network reliability. While prior research has examined station-level and system-wide crowding, little attention has been given to quantifying how individual stations contribute to link-level overcrowding. This study addresses this gap by introducing the Crowding Contribution Index (CCI), a metric that quantifies the extent to which destination stations drive overcapacity flows on preceding links. The CCI is computed via a structured framework integrating Automated Fare Collection (AFC) and GTFS link-network data. Applied to over 80 million trips across 237 Delhi Metro stations, 142 200 hourly CCI values reveal that 46.35% of station-hours exceed capacity, with highest contributions clustered in specific stations. A Type II Tobit model assesses built-environment (BE) variables, showing that POI and intersection densities increase contributions, while POI entropy reduces them, underscoring land-use diversity’s role. Random Forest and XGBoost models corroborate these findings, ranking BE variables as the strongest CCI predictors. These insights emphasise the need for integrated land-use and transport strategies. The CCI framework offers operators a scalable tool for real-time service adjustments, such as targeted short-turns and dynamic fleet deployment, and guides planners toward sustainable, integrated land-use planning, making it especially valuable for rapidly urbanising, data-constrained cities.