基于呼叫详细记录数据的城市居民职业属性提取模型

Occupational attribute extraction model for urban residents based on call detail record data

了解城市内居民的社会属性,如职业属性,对公共政策的制定具有重要意义。很多研究通过分析社交网络行为来获取人员社会属性信息,却忽视了社会属性信息与个人日常活动之间的关系;还有部分研究使用时空数据进行分析,但受到样本特征的限制。呼叫详细记录(CDR)间接记录了粗糙的人员日常活动轨迹,基于CDR提出城市居民职业属性提取模型,先提取居民停留点和轨迹链,再计算职住位置,并结合兴趣点(POI)信息构建个人出行特征和区域特征,最后使用半监督分类模型提取城市居民职业属性。对覆盖600万人的CDR数据来分析和提取两种职业类型的人:学生和城市蜂鸟,最终模型的F1得分超过0.95。

Understanding the social attributes of urban residents,such as occupational information,is of great significance to the formulation of public policies.Many researches obtain the social attributes of people by analyzing social networks,but ignore the relationship between social attributes and personal daily activities.Some studies use spatio-temporal data for analysis,but are limited by sample characteristics.Call Detail Records(CDR)indirectly record rough trajectories of people’s daily activities.Based on CDR,an occupational attribute extraction model for urban residents was proposed.The stopping points and trajectory chains of residents were first extracted,then their occupational locations were calculated,and the personal travel characteristics and area characteristics were constructed by combining Point Of Interest(POI)information.Finally occupational attributes of city residents were extracted by using semi-supervised classification models.The CDR data covering 6 million people was used to analyze and extract two occupational types:students and urban hummingbirds.The final model’s F1 score exceeded 0.95.