Developing an area classification system from mobile phone usage data

Abstract

Since human activities are vary by time and place, there have been many attempts to extract social behavior in spatial studies which included area use in the city. This information helps gaining advantages in city and facilities planning. Nowadays, many people carry mobile phone with them for communication purpose. This motivates us to analyze mobile phone usage in different area types. This thesis proposes method of an analysis for area use classification from mobile phone usage pattern. CDR data was used to define mobile phone usage pattern by hour from 1:00 to 24:00 and day of week from Monday to Sunday. And then processed clustering algorithm to find optimal groups of usage pattern from cell towers. Ratio of usage from each pattern is used to predict area type and facilities. Besides, the researcher also revealed the examples of mismatch location and minor groups of area use in this thesis.