摘要
应用C5.0决策树算法、CART决策树算法、RBF神经网络建立某固网运营商客户的收入流失预警模型,然后运用GMDH方法建立了客户的收入流失组合预警模型。研究表明:模型提高了预警的准确率和覆盖率,模型能够帮助运营商找出收入有可能流失的客户,使其能对这些客户开展提前、有效的挽回营销工作,最大程度减少客户流失。
This paper uses the C5.0,CART and RBF neural network to establish early warning model of wireline telecom customer lose respectively;and then uses GMDH algorithm to make a combined prediction model.Model improves the accuracy and coverage of early warning.It can help operators find the customers who will defect in advance,and implement earlier and more effective marketing measures to reduce the customer lose.
出处
《软科学》
CSSCI
北大核心
2012年第1期128-131,共4页
Soft Science
基金
国家自然科学基金资助项目(70771067)
关键词
数据挖掘
客户流失
组合预警
data mining
customer lose
combined early warning