摘要
[目的/意义]旨在根据用户细分结果进行用户流失预测。[方法/过程]采用聚类分析方法对微博社区功能结构及用户特征进行分析,对用户流失条件进行界定,并进行用户分类;采用典型判别分析法构建用户流失预测模型,并通过交叉验证法判别预测精度。[结果/结论]该用户流失预测模型是可行的,且精度较高。提出"活跃型用户"和"明星用户"应加强回访和特权服务;"信息索取型用户"和"沉默型用户"应采取信息推送、Email等方式激发其活跃度;应建立用户流失预警机制,定期进行用户流失预测分析等建议。
[Purpose/significance]The paper is to carry out a customer churn prediction according to customer subdivision results.[Method/process]The paper analyzes the functional structure and users’feature of the Microblog community by clustering analysis,defines the customer churn conditions,and categorizes the users.It establishes the customer churn prediction model based on typical discriminant analysis,and distinguishes the prediction accuracy by cross-validation.[Result/conclusion]The customer churn prediction model is feasible,and its accuracy is quite high.It puts forwards countermeasures,such as strengthening return visits and privilege service for active customers and star customers;stimulating vitality of information requiring customers and silent customers by information push and Email;establishing customer churn early warning mechanism,and making a regular prediction.
作者
贺芳
He Fang(Library of Zhongnan University of Economics and Law,Wuhan Hubei 430073)
出处
《情报探索》
2018年第12期21-27,共7页
Information Research
关键词
微博用户
用户细分
用户流失预测
Microblog users
customer subdivision
customer churn prediction
