Clustering Mobile Call Detail Records to Find Usage Groups

Clustering techniques, applied to anonymized cellular call detail records from a small city, reveal groups with similar usage patterns. Voice and text message usage, attributed to a particular day of the week and hour of the day, was processed to produce seven compact and well separated clusters. Two of these are investigated in detail and are related to: a) commuters who call on the way to and from work; b) students who have text message patterns related to the daily school schedule and are handled by cell sites that cover the local high-school and middle-school.