Identification and Characterization of Human Behavior Patterns from Mobile Phone Data

The availability of datasets coming from the telecommuni- cations industry, and specifically those relevant to the use of mobile phones, are helping to conduct studies on the pat- terns that appear at large scales, and to better understand social behaviors. This study aims at developing methods for enabling the extraction and characterization of normal be- havior patterns, and the identification of exceptional, or di- vergent behaviors. We study call activity and mobility pat- terns to classify the observed behaviors that exhibit similar characteristics, and we analyze and characterize the anoma- lous behaviors. Moreover, we link the identified behaviors to important events (e.g., national and religious holidays) that took place in the same time period, and examine the interplay between the behaviors we observe and the nature of these events. The results of our work could be used for early identification of exceptional situations, monitoring the effects of important events in large areas, urban and trans- portation planning, and others.