Integrating six feeder modes with the metro system: Insights from on-demand and pooled services

Abstract

Understanding travel patterns of feeder modes for metro systems is crucial for addressing first-mile challenges and advancing sustainable urban mobility. Existing research predominantly examines traditional transit modes (e.g., buses) and emerging on-demand transport services (e.g., ride-hailing) separately, with limited comparative analysis between the two. Furthermore, most studies do not distinguish between solo trips and pooled trips, leading to potential misestimation of the impact of each trip type. Drawing on multi-source data from Shenzhen, this study analyzes the access patterns of six feeder modes (i.e., solo ride-hailing, ride-pooling, solo taxis, taxi-pooling, buses, and free-floating bike-sharing) and their integration with the metro system. We employ the eXtreme Gradient Boosting (XGBoost) and SHapley Additive exPlanations (SHAP) models to evaluate the impacts of these feeder modes on metro ridership, and analyze how on-demand transport services fill gaps in the metro-bus network based on the coverage extension ratio and travel time ratio. Our findings reveal: (1) Solo taxis represent the fastest feeder mode; (2) As access distance increases, the cost advantage of ride-hailing diminishes, while the speed advantage of taxis becomes more pronounced; (3) Ride-pooling contributes to increased metro usage in non-central areas; (4) On-demand transport services can effectively address gaps in the metro-bus network. This study provides policy implications for optimizing transportation resource allocation and enhancing first-mile connectivity for metro systems.