计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (9): 154-156.

• 网络、通信与安全 • 上一篇    下一篇

数据挖掘方法在用户流失预测分析中的应用

刘光远 苑森淼 董立岩   

  1. 吉林大学通信学院
  • 收稿日期:2007-01-15 修回日期:1900-01-01 出版日期:2007-03-21 发布日期:2007-03-21
  • 通讯作者: 刘光远

Prediction the Churn of Customers with Data Mining Method

  • Received:2007-01-15 Revised:1900-01-01 Online:2007-03-21 Published:2007-03-21

摘要: 摘要:移动通信在高速发展的同时,也出现了大量用户离网的问题,如何基于客户信息、消费行为等历史数据,预测客户离网的倾向是一项重要的课题。面对庞大的原始数据集,数据挖掘在这个领域中发挥了重大作用,本文基于客户的历史数据和短期偶发数据,提出了链型挖掘的方法,并结合决策树,形成了一个综合的链型树分类器(Chain Tree Classifier, CTC),对CTC分类器进行了实验仿真,实验结果显示,CTC分类器对移动通信运营商感兴趣的单个事件发生的顺序进行了良好的预测,可以从中得到客户离网可能性的概率,从而帮助运营商寻求相应的方法来留住客户,降低离网率。

关键词: 数据挖掘, 链型树分类器

Abstract: Abstract:In the telecommunication, a major topic is to predict the churn of customers based on the database with customers’ information and calling history. Data Mining plays an important role in the prediction of churn. This paper proposed a chain data mining method to use the historical and temporary data and combined with a Tree database structure to form a Chain Tree Classifier (CTC). Simulation with a group of data shows a satisfied prediction result for the events what the industry of telecom is interested. The method could help the companies to find the corresponding methods to keep the customers and obtain the good revenue.

Key words: Data Mining, Chain Tree Classifier (CTC)