[2C5-OS-7b] OS-7 (2)

[2C5-OS-7b-01] A study of user profiling using location data with machine learning

〇Yutaro Mishima¹, Rui Kimura¹, Atsunori Minamikawa^2,1 (1. KDDI Research, Inc., 2. KDDI Corporation)

Keywords:Location Data, User Profile, Machine Learning, Targeted Advertising

These days, social issues like air pollution and aging become so serious that smart city will be needed in the near future. According to this change, it is important to know profiles of people accurately. For example, accurate information on age distribution, family structure and rates of car ownership will lead optimizing child monitoring services and MaaS services like car-sharing. However, there is no measure to obtain accurate profile mentioned earlier. Although we can obtain age distribution and family structure of cities from a national census, they are updated only once every five years. To deal with these issues, we are looking forward to the method by which we can estimate users’ profile from location data of their smartphones accurately and we have an experiment as primary research by using location data and a questionnaire. As a result of experiments, we indicate the method significantly enhances the accuracy of estimation in some profiles by processing location data into useful features.