数据挖掘在移动通信业大客户离网预测中的应用  

Using Data Mining to Build Churn Prediction Model for High-Level Customers in Mobile Communication Market

作　　者：王姝华[1] 钟云飞 

机构地区：[1]江苏移动通信有限责任公司信息技术中心，江苏南京210029 [2]北京瑞斯泰得数据技术开发有限公司技术部，北京100086

出　　处：《江苏通信技术》2004年第3期1-4,共4页Jiangsu Communication Technology

摘　　要：大客户是各移动运营商利润的主要来源，也是竞争的焦点。随着市场竞争的日益激烈，如何降低大客户离网率，是摆在各运营商面前的战略性任务。采用数据挖掘技术，遵循数据挖掘标准流程CRISP-DM。从商业理解、数据理解、数据准备、建立模型、模型评估和结果部署等6个阶段，详细介绍了移动通信企业中大客户离网预测模型的建立过程和方法。同时对预测结果从技术和业务上进行深入分析，以辅助运营商及时采取措施进行挽留。High-Level customers are the main source of profit of every mobile communication provider, and are the focus of the competition between these providers. With the increasing of the competitive market, how to control the customer leaving is a strategic task for the providers. According to the standard data mining process, CRISP-DM, the procedure of a high-level customer churn prediction model is discussed thoroughly from the six phases of business understanding, data understanding, data preparation, modeling, evaluation and deployment. And more the predictive results from the technology and business view are deeply analyzed in order to help reduce the high-level customer leaving.

关 键 词：数据挖掘 大客户离网 预测模型 移动通信业 CRISP-DM 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TN929.5[自动化与计算机技术—计算机科学与技术]