基于收视行为模式的有线电视用户流失预测  

Prediction of Cable TV Customer Churn Based on Viewing Behavior Mode

作　　者：万倩[1] 赵明[1] 王炳飞 

机构地区：[1]国家新闻出版广电总局广播科学研究院,北京100866 [2]北京数码视讯科技股份有限公司,北京100085

出　　处：《广播与电视技术》2016年第11期67-71,共5页Radio & TV Broadcast Engineering

摘　　要：面对激烈的市场竞争,广电运营商一方面需要提升服务质量,吸引新入网用户,同时也需要更加重视对用户流失的预测管理。利用数据挖掘技术进行用户流失预测在有线电视业务中的应用尚处于探索阶段,本文针对有线电视业务的特点,提出了基于RFM模型的流失用户定义,对用户的属性特征进行分析,得出流失用户的基本特征,并提出基于用户的收视行为模式进行流失预测的分析算法,最后通过实际数据进行仿真实验,验证了模型和算法的可行性。Facing fierce market competition, radio and television operators need to upgrade quality of services to attract new customers.They also need to pay more attention to customer churn prediction. Using data mining techniques to predict customer churn in the cable television business is still in an exploratory stage. According to the characteristics of cable TV business, this paper proposes the definition of churn customer based on RFM model, analyzes customer attributes, gets essential features of lost customers, and proposes churn prediction analysis algorithm based on the viewing behavior mode of customers. Finally, we carry out simulation experiment with the actual data to verify feasibility of the model and algorithm.

关 键 词：数据挖掘 用户流失预测 决策树 有线电视 

分 类 号：TN94[电子电信—信号与信息处理]