Transport Mode Choice for Disaggregated Mobility Demand Generation

The OpenStreetMap Mobility Demand Generator (OMOD) is an open-source tool that creates realistic synthetic mobility demand for any region in Germany. The tool focuses on ease of use and fast prototyping, requiring only OpenStreetMap (OSM) data from the user. In this work, we introduce the aspect of traffic mode choice into OMOD utilizing a two-stage multinomial logit model. The model works with the precise coordinates of the trip origins and destinations and exact departure times to determine each trip’s travel time and other level-of-service attributes based on network information acquired from OpenStreetMap and public transit schedules in the form of GTFS files. We evaluate the performance of our model in two stages: first, directly on a separate test set of the household travel survey, and second, for a microsimulation of three different German cities. On the household travel survey, we reach accuracies between 64.3 % and 77.7 % percent with well-fitted calibration curves. Achieving high performance in the microsimulation is more difficult due to the existence of downstream errors. Nonetheless, we achieve close fits for all modes with the exception of the car passenger mode.