Inferring Origin-Destination Demand and User Preferences in a Multi-modal Travel Environment Using Automated Fare Collection Data

Abstract Eliciting individual travelers’ Origin-Destination (OD) information is critical for enabling public transit system policy-makers and operators to serve travelers in a calculated way. Accurate estimation of route choice model parameters is also important, in that it can help assess or predict the service levels that such a system can be expected to achieve. The knowledge of both the OD links and route choice logic is especially in demand for emerging mobility services, where providers work to accommodate individualized services and also offer incentives to travelers for specific trips. We show that all this information can be distilled from a particular type of data – the Automated Fare Collection (AFC) system data – in a fast, low-cost way. This paper presents a two-step methodological framework to identify individual travelers’ true ODs (beyond stop-level ODs), as well as infer their travel preferences. The key to our work is the ability to identify and process the observations of travelers’ routing choices between the same ODs under different travel environment conditions. A presented specially-crafted case study validates the proposed method in application with a real-world AFC data of Seoul, Korea, confirming the method’s high inferential ability, under a basic route choice model.