基于数据挖掘的呼叫中心IVR客户细分研究  

Customer segmentation of call center IVR based on data mining

作　　者：莫赞[1] 樊鹏[1] 李燕飞[1] 

机构地区：[1]广东工业大学管理学院,广东广州510090

出　　处：《微型机与应用》2014年第12期61-63,共3页Microcomputer & Its Applications

基　　金：国家自然科学基金项目(71171062);"十二五"国家科技支撑计划项目(2011BAD13B11);广东省哲学社科规划(GD10YGL09);广东工业大学教学质量工程项目(402102283)

摘　　要：语音应答业务利用率低、人力需求成本高是一大问题,对此,结合案例,运用CRISP-DW方法,采用Clementine工具对所得数据进行分析处理,得到IVR客户与地区分布、ARPU值和入网时长的关系;对IVR客户进行聚类分析,总结出其重要特点;提出相应对策,帮助提高IVR客户的满意度和利用率。建立一个C5.0决策树模型,分析预测客户是基于什么情况才进入IVR系统的。分析IVR客户与入网时长、地区、ARPU值之间的规则,有助于更准确地了解IVR客户并对其细分。Voice response service （ IVR ） has low utilization , high cost of manpower needs . So , this paper combines with the actual projects , and makes reference to CRISP-DM data mining methodology and uses Clementine data mining tools to analyze the data . In addition , it pre-treats the operation data of call center . Based on data arrangement and pretreatment , it analyzes the relationship between IVR customers with regional distribution , ARPU value , and in_months , and then makes cluster analysis on IVR customers . Through clustering , it analyzes the relevant features of IVR customers and puts forward measures to improve the IVR customer satisfaction and utilization . It establishes a C5 . 0 decision tree model to analyze the model to predict the customers enter the IVR system . It analyzes the rules between IVR customers and the three data items of regional distribution , ARPU val-ue , and in_months .

关 键 词：语音应答业务 数据挖掘 CRISP—DM 客户细分 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]