A Social Attribute Inferred Model Based on Spatio-Temporal Data

Understanding the social attributes of urban residents, such as occupations, settlement characteristics etc., has important significance in social research, public policy formulation and business. Most of the current methods for obtaining people's social attributes by analyzing of social networks cannot reflect the relationship between the occupational characteristics and their daily movements. However, the current methods of using spatio-temporal data analysis are limited by the characteristics of the samples, and focus more on travel patterns and arrival time predictions. Based on coarse-grained CDR (Call Detail Record) data, this paper proposes an approach to infer occupation attribute by analyzing the travel patterns of personnel and incorporating more enhanced information. Finally we uses the CDR data of 6 million people to analyze and extract two types of people: college students in Beijing and urban hummingbirds and the F1 score of our proposed model is more than 0.95.