摘要
用户离网预测对于电信企业维系和挽留高价值用户具有非常重要的意义。通过对用户离网预测问题的分析,指出构建用户离网模型的关键因素在于业务理解和数据挖掘算法选择。近年来各类机器学习算法已经被大量应用到电信企业数据挖掘实践中,通过介绍并对比4种经典的机器学习算法,指出在标准化输入数据之外,选择合适的数据挖掘方法,并有技巧地合并不同用户离网预测模型的预测结果,可以显著地提高用户离网预测成功率,最后给出了应用示例。
For telecom operators, customer churn prediction is important to retain valuable users. Through the analysis of the problem of customer churn prediction, it points out that that the key factors of the customer churn analysis model are business under- standing and data mining algorithm selection. In recent years, many machine learning algorithms have shown their applicability to the data mining problems raised in telecom operators. By introducing and comparing 4 classical machine learning algo- rithms, It points out that in addition to standardized input data,selecting appropriate data mining algorithms and skillfully combining the prediction results of different customer churn prediction models can significantly improve the success rate of churn prediction. Finally,an experimental example is given.
作者
董润莎
徐争莉
袁明强
程新洲
Dong Runsha;Xu Zhengli;Yuan Mingqiang;Cheng Xinzhou(China Unicom Network Technology Research Institute,Beijing 100048,China)
出处
《邮电设计技术》
2018年第10期1-5,共5页
Designing Techniques of Posts and Telecommunications
关键词
用户离网
电信企业
机器学习
Customer churn
Telecom operators
Machine learning
