Mobile Communication User Profiling Based On Call Detail Records

In recent years,it is a hot topic in academia and industry to mine user attribute features based on user data and user profiling is an important basic research point.User profiling can understand the needs of individuals,tap the characteristics of group users,and discover potential user groups in order to provide users with better services and personalized recommendations.Since China entered the age of mobile communications.the connection between people and mobile phones has become closer,and mobile phones have become an integral part of people's daily lives.The mobile phone generates historical data that records the user's location information,usage information,and social information.It is a valuable data resource,which implies a lot of personalized information.and provides a new information channel for studying the attributes of users and grasping the pulse of urban life.In this thesis,the study of individual portraits and group portraits of mobile communication users is completed through the analysis of call detail records,which provides a new perspective for the inference of user attributes and the understanding of user characteristics,and also provides references for the improvement of production and living.and the formulation of relevant policies.Currently.the research of call detail records is mostly focused on the spatio-temporal information.We comprehensively use the spatio-temporal information.the social information and the law of active time to study the portrait of the user.First of all,in order to analyze the user's spatio-temporal data more conveniently,we use the relative positioning method between points to convert the GPS latitude and longitude coordinates into the common two-dimensional plane rectangular coordinates,which makes the calculation between communication cells more convenient and preserves the distance relationship and azimuth relationship.Secondly,to quantify the user accurately and completely,we extract the moving distance,the radius of gyration,the number of access points,the entropy of moving directions to characterize user's mobile pattern and extract the call duration,the number of contact.the ratio of calling,the entropy of sociality to characterize user's social life and extract the user's active time regular feature.And then,this thesis processes the user's features in different rules and generates user tags in order to be clearly presented by the characteristics of the user's individual.-Then we use wordle tools to generate a word cloud card for each user.Finally,users are divided into four groups according to the mobile pattern and social life.We explore the characteristics of various user groups and analyze the regularity of user groups in terms of active time.This thesis also explores the relationship between user movement patterns.social interaction and user activation time so as to deepen our understanding of user characteristics.