Uncovering mobility typologies of territorial zones based on Floating Car Data mining

This paper describes a data exploration study using Floating Car Data to analyze mobility patterns of geographical spaces. The objective is to build a mobility-related typology of territorial zones by investigating related vehicle movements. Mobility features at the level of trips and stay places are recovered from daily vehicle trajectories. Place visiting frequentation is further analyzed at each vehicle level to identify significant places and corresponding activity regularity. Based on these mined patterns, a multi-view cooperative clustering method is developed to feature out the zonal mobility typology in terms of the composition of local stays, temporal flows of trip generation and attraction, and spatial connections in trip distance distribution. The proposed framework was applied to the Great Paris region for an experiment using 14 days data. Consequentially, 5 mobility types of zones were obtained, with each holding a different orientation of mobility usage. Discovered areas were also compared with the common recognition of their social functions, which showed a consistent matching. Overall, this study provides a data-driven approach to study mobility interactions with territorial spaces, by spatial segmentation, characterization, and differentiation.