Enhanced recursive logit modeling for intermodal travel by integrating path-level attributes: A data-enriched application using smartcard and survey data

Understanding intermodal travel behavior is crucial for data-driven transportation policy and planning. This paper develops enhanced Recursive Logit and Nested Recursive Logit models that incorporate path-level attributes, such as travel costs, to comprehensively analyze intermodal traveler choices. To achieve this, we propose a methodology that represents path-level attributes at the link level, minimizing errors by using the pseudoinverse of the path-link incidence matrix and enabling their incorporation into link-based utility functions. By incorporating intermodal path cost as a path-level attribute within the Recursive Logit modeling frameworks, the proposed models capture the effects of cost on mode and path choices, and enable the identification of key policy indicators such as the Value of Link Travel Time (VoLTT) and the Value of Transfer (VoTR) between different transportation modes. Integrating transit smartcard and travel survey data from Seoul, South Korea, we construct an empirical dataset of intermodal trips with specific trajectories to estimate and cross-validate the models. Results demonstrate that the proposed models outperform traditional approaches in capturing real-world intermodal travel behavior. By providing data-driven insights into the values of link travel time and transfers, this research supports optimizing intermodal transportation systems and promoting sustainable mobility strategies.