Identifying human spatio-temporal activity patterns from mobile-phone traces

Understanding and modeling people´s mobility is a crucial component of transportation planning and management. Research in this area was originally concentrated on modeling commuting flows as they generally account for a vast majority of trips. Nowadays however, more and more trips are done to perform other activities, such as leisure. Identifying the types of places visited during a trip can be beneficial to understand the performed activities and so characterize the daily mobility of a population. In this paper we analyze a large mobile phone location dataset to monitor human locations over the course of two week time interval. We then map human locations to geographical features of the visited places and use that to characterize the daily human mobility. A limited number of visited land use patterns is found that allows describing different types of people and their daily mobility choices. The resulting patterns are characterized with peculiar trip lengths and home locations, thus showing interesting insights into modeling human travel demand, with applications to transportation activity-based models and place recommender systems.