The mobility patterns of the population are the basis of most analyses in the transportation field. We aim to extract these patterns from smartphone traces. The following thesis proposes a Bayesian approach based on smartphone location records, land use information and schedule data to understand the activities that people daily perform. We investigate two alternatives. The prior is either based on schedule data or based on land use information. We test the algorithm on the smartphone WiFi traces provided by Nokia. They have been obtained from people who live around the Leman Lake, essentially in the region of Lausanne. The Swiss Federal Statistical Office (FSO) and OpenStreerMap (OSM) provide the schedule and the land use information. The results show that the prior based on schedule data is not convenient: the importance of the individual behavior decreases, and the activity Home comes up too often. The results are much better when the prior depends on the land use that surrounds a users’ location of interest. In addition, we have successfully extracted recognizable mobility patterns, particularly for the activities Home and Work.