A Seriation Based Framework to Visualize Multiple Aspects of Road Transport from GPS Trajectories

Dubray, Alexandre [UCL] Nijssen, Siegfried [UCL] Thomas, Isabelle [UCL] Schaus, Pierre [UCL]

Heavy good vehicles GPS trajectories can tell us a lot about the structure of the road network and the spatial organization of a territory. In this paper we introduce a framework for visualizing multiple aspects of road transport on a single map. The final result allows unveiling at one glance areas with similar properties and lets the user gain insight in the geography of a studied area. More concretely, the map is divided into basic spatial units and for each spatial unit relevant features are computed from the GPS trajectories, such as the time spent in congestion, or resting. A seriation algorithm is then used to discover a linear order of the spatial units where successive units of the order share similar features. A rainbow color scale is used to assign a color to each spatial units to visually reveals similarities with respect to the extracted features. As a case study, we analyze the Belgian territory using a real-world data set. We compare our approach with clustering based methods, demonstrating that it reveals more of the spatial structure. The source code of our implementation is available under an open-source license for anyone interested in analyzing GPS trajectories.