Learning from Experts: Inferring Road Popularity from GPS Trajectories

Recent years have witnessed a steep increase in the collection of movement data through GPS trajectories. Such data sets have great potential for providing insights into mobility demand and behaviour for city planners, or to improve routing services for the end user. We propose a method for inferring popularity from GPS trajectories. The inferred popularity is assigned to a road graph, and is suitable for routing with Dijkstra´s algorithm. The inference method can be calibrated with several parameters. In this paper we describe the inference method, demonstrate the influence of the available parameters on a data set of cycling trips in the city of Vienna, Austria, and compare popularity routes to routes optimized for other criteria.