Modeling Transit Travel Patterns from Location-Stamped Smart Card Data Using a Disaggregate Approach

Data collected by Global Positioning System (GPS) based smart card fare collection system adopted by numerous agencies cannot readily be used for service planning purposes due to erroneous and incomplete travel data. Based on a database with more than 750,000 smart card boarding transaction records per month from a transit agency in the National Capital Region of Canada, this paper presents a multi-step method to develop a complete procedure which identifies and imputes incorrect or suspect data, then derives the most probable alighting bus stops and finally provides suitable origin-destination travel data necessary for the application of transit trip assignment algorithms.