节点文献

数据挖掘技术在客户流失预警中的应用

Application of Data Mining Technology in Customer Churn Prediction

【作者】 崔永哲

【导师】 崔荣一;

【作者基本信息】 延边大学 , 计算机应用技术, 2008, 硕士

【摘要】 数据挖掘是从大量的数据中抽取出潜在的、不为人知的有用信息、模式和趋势。其目的是提高市场决策能力、检测异常模式、在过去的经验基础上预言未来趋势等等。它致力于数据分析和理解、揭示数据内部蕴藏知识的技术,已成为未来信息技术应用的重要目标之一。本学位论文针对通信行业小灵通客户流失问题进行了研究与分析,以C4.5决策树数据挖掘算法为核心设计实现了客户流失预警模型并予以了实际应用。首先,以决策树数据挖掘算法为理论基础,结合ID3算法和C4.5算法对训练样本集合进行实例分析和对比,确定了建立客户流失预警模型的算法;其次,针对通信行业小灵通客户流失问题,以C4.5算法为核心设计了预警模型;最后,从延边网通分公司现有运营支撑系统提取客户数据作为数据源,对数据进行了清洗和预处理,利用本文设计实现的客户流失预警模型对清洗后的客户数据进行分类、评估,挖掘出流失用户的业务特征并生成客户流失分类规则。应用结果表明,本文所实现的客户流失预警系统可为延边网通分公司提供较准确的决策依据。据此,公司可以对流失倾向较高的客户群体采取有针对性的客户挽留策略,很大程度上避免了因客户流失对企业造成的损失,为延边网通分公司更好地开展客服工作起到了积极的推进作用。在理论知识商业化应用方面,本学位论文进行了一次有意义的探索和尝试。

【Abstract】 Data mining means to take out the potential unknown information mode and trend from the data, in order to improve the ability of market-decision and abnormity-test mode, to predict the tend in future on the base of experience in the past. It devotes to digital analysis and understanding, to find out potential technology in the data. It has become major object on applying of information and technology.In this dissertation churn prediction the Decision Tree Algorithm of data mining is researched which is applied the customer churn prediction. Firstly, the basic concept of Data mining is introduced, and several classification methods of decision tree are elaborated in detail, including ID3 algorithmic method which means dividing nodes on entropy attribute, C4.5 algorithmic method which could deal with the continuous attribute and absent value. Then an improved method of the C4.5 is proposed and its characteristics is researched. Customer churn prediction of the PHS (Personal Handy-Phone System), resulting of the serious competition in the telecom field, is an expression problem with practical meaning. The decision tree is applied to analyze the customer churn of the PHS in the field of telecommunication in this dissertation, including building mode, issue, compare, analysis, and reasoning conclusion with SAS. It has been carried on a deep exploration and attempt in this dissertation on the commercialized application of the theoretical knowledge.

  • 【网络出版投稿人】 延边大学
  • 【网络出版年期】2010年 01期
节点文献中: 

本文链接的文献网络图示:

本文的引文网络