The recent Swedish Intelligent Speed Adaptation (ISA) study included a component that involved the installation of units based on the Global Positioning System (GPS) in hundreds of cars in three Swedish cities, Borlänge, Lund, and Lidköping; these vehicles were observed for up to 2 years. In Borlänge, the speed and location data of each vehicle were transmitted at regular intervals to a central server and stored for later analysis. This data set contains a wealth of travel behavior information that had not been available before. However, a data set of this magnitude introduces a major need for automated processes that can glean travel behavior details from the trip summary and collected GPS point files. A summary is presented of characteristics of and issues with the Borlänge GPS data set, which included 186 personal vehicles with at least 30 days of travel data and corresponding household sociodemographic data. (These 186 vehicles recorded 49,667 vehicle days of travel and 240,435 trips inside the study area.) Then, automated methodologies are presented for imputing trip purpose for these trips once the trip destinations are identified, as well as for correcting the GPS traces and identifying missing trip ends within these trips. Results of these automated processes for a subset of the ISA study vehicles are included.

No abstract