Discovering vehicle usage patterns on the basis of daily mobility profiles derived from floating car data

ABSTRACT

This paper presents a novel approach for establishing vehicle usage patterns by Floating Car Data based on their daily mobility making. Vehicle trajectories were firstly sequenced into meaningful trips and then clustered into different trip types. Trips pertaining to each vehicle were aggregated as a vector of counts per type to obtain the mobility profile of the vehicle. Based on these profiles, a topic modeling approach using Latent Dirichlet Allocation was developed to discover the patterns of vehicle daily usage, thereby constituting a typology. An application was conducted for the Paris Region, which identified 3 vehicle usage types associated with local usage within specific areas and the other two holding hybrid patterns between different areas. The prevailing pattern of vehicle usage was found on short-medium trips around pericenter and near suburban areas. Overall, this study offered a data-driven framework to help understand vehicle daily usage patterns and their differentiation

KEYWORDS:

• Mobility pattern
• vehicle daily usage
• trajectory segmentation
• Floating Car Data
• machine learning

Acknowledgments

The research was supported by Chair Eco-design Vinci at Ecole des Ponts ParisTech. The authors would like to thank Geolytics project, with partners IT4PME and Coyote, for the data and assistances, and Chair ENPC-IdFM in facilitating this research.

Disclosure statement

No potential conflict of interest was reported by the authors.