面向零售业客户关系管理的数据挖掘模型研究

【作者】 汪毅；

【导师】 朱顺泉；

【作者基本信息】 湖南大学 ， 管理科学与工程， 2003， 硕士

【摘要】 从20世纪90年代起，随着市场竞争的日益激烈和经济全球化的趋势，也由于不断发展的信息技术较好地解决了战术性的日常运作事务问题，许多企业家和学者开始把关注的焦点从改进内部运作更多地放到管理客户资源上来，并因此提出了所谓客户关系管理(CRM,Customer Relationship Management)的概念。人们普遍认为，通过客户关系管理，企业能更好地为客户服务，最大限度地保持和发展客户，进而提高企业的竞争力。客户关系管理的实现，可以从两个方面考虑。一是从管理的角度，解决管理理念问题；二是从技术的角度，为这种新的管理模式提供信息技术的支持，而其中的核心就是数据挖掘技术。由于已经有不少从管理角度讨论客户关系管理的研究成果，所以本文没有过多地讨论客户关系管理的管理理念，而是以零售业为例，在介绍客户关系管理基本理论和其他相关技术的基础上，紧密地围绕其核心技术——数据挖掘进行研究和论述。 论文首先从客户关系管理和数据挖掘的基本概念和原理入手，阐明了在客户关系管理中应用数据挖掘的必要性，构建了以数据挖掘为核心的CRM系统框架；然后，论文改进和扩展了传统的关联规则和决策分类树模型，提出了具有确定性时间约束的关联规则和模糊决策分类树，并修改了传统的挖掘算法，通过示例展示了新模型的应用方法。更多还原

【Abstract】 From 1990’s, with the increasing vehemence of market competition and the current of economic globalization, because the trivial affair now can be well done by continually developing information technology, many enterprisers and scholars begin to put the more attention to the management of customer resources than the improvement of inner management. And then, they put forward the concept of customer relationship management. We generally think that if a corporation implements CRM, it can serve its customers better, hold and expand customers furthest, and improve the power of competition. The realization of CRM can be thought in two aspects. One aspect is management for settling the problem of managing idea, the other is technology to provide technical support to the new management modal, and data mining (DM) is the core of this support. Many research fruits on CRM are based on the aspect of management. So, this thesis didn’t discuss the management idea of CRM. It researched and discussed CRM through DM closely at the basis of introducing basic theories of CRM and other relative technologies with the example of retailing.At first, this thesis described the necessary of implementing DM in CRM systems at on the basis of explaining the elementary concepts and principles of CRM and DM, constructed a CRM system framework with the center of DM. Then, it ameliorated and expanded the models of traditional association rule and decision tree for classification, put forward association rule with time constraint and fuzzy decision tree for classification. The thesis amended traditional algorithms and showed the application methods of new models.更多还原