Clustering par optimisation de la modularité pour trajectoires d'objets mobiles

Mohamed Khalil El Mahrsi (1) , Fabrice Rossi (2)

1 LTCI - Laboratoire Traitement et Communication de l'Information
2 SAMM - Statistique, Analyse et Modélisation Multidisciplinaire (SAmos-Marin Mersenne)

Analysis and data mining of moving objects trajectories have gained a considerable amount of interest in the last few years. In this article, we present a clustering approach tailored for trajectories of vehicles moving on a road network. First, we introduce a similarity measure that makes it possible to compare such trajectories while taking into account the constraints of the underlying network. Then, this measure is used to construct a graph that models the interactions among the trajectories w.r.t. their similarity. A community detection algorithm based on modularity optimization is applied to the graph in order to discover groups of trajectories that behaved similarly and that moved along the same portions of the road network. We implemented the proposed approach and tested it on multiple synthetic datasets in order to show its feasibility and its efficiency.

Mots clés: trajectoires; objets mobiles; similarité; classification

Domaines: Machine Learning [stat.ML]; Apprentissage [cs.LG]; Informatique ubiquitaire

Mohamed Khalil El Mahrsi, Fabrice Rossi. Clustering par optimisation de la modularité pour trajectoires d'objets mobiles. Ubimob'12, Jun 2012, Anglet, France. pp.12-22. ⟨hal-00710657⟩