Constructing a routable retrospective transit timetable from a real-time vehicle location feed and GTFS

We describe a method for retroactively improving the accuracy of a General Transit Feed Specification (GTFS) package by using a real-time vehicle location data set provided by the transit agency. Once modified, the GTFS package contains the observed rather than the scheduled transit operations and can be used in research assessing network performance, reliability and accessibility. We offer a case study using data from the Toronto Transit Commission and find that substantial aggregate accessibility differences exist between scheduled and observed services. This 'error' in the scheduled GTFS data may have implications for many types of measurements commonly derived from GTFS data.