【作者】 支永安；

【导师】 欧阳一鸣；

【作者基本信息】 合肥工业大学 ， 计算机系统结构， 2010， 硕士

【摘要】 在中国电信运营商重组以后,国内电信业的市场环境已渐趋合理,它们之间的竞争更加激烈。客户是电信运营商生存和发展的根基[1],对客户的保有和争夺也就成为竞争的焦点。作为全业务运营商,中国电信在C网业务发展的同时,导致部分固网业务的下滑,同时由于传统的固网是后付费方式,C网的用户大多还未接受预付费方式,这给企业带来了极大的欠费风险和收入增长压力。为兼顾客户感知和企业利益,必须实现差异化服务和精确化营销,解决以往在固网中应用的普遍催缴的欠费管理方式,引入银行业的授信方法,提高企业的品牌形象和有针对性的欠费管理模式。本文主要研究内容是在数据仓库的基础上,对用户的日消费行为进行分析和挖掘,获得内在规律性,实现对客户的差异服务和管理。在电信业,普遍存在着“信息丰富”,“知识贫乏”的现象,传统处理方法难以发现其中蕴含的知识,迫切需要新知识、新技术来解决这个问题。用数据仓库对信息进行管理,并在此基础上对信息进行挖掘,从而应用于客户管理,销售策化,资源配置等诸多方面。在现实生产环境中,信息获取的高效和业务的智能,是挖掘结果能否被应用推广的至关重要因素。为此,本文在数据仓库的设计和组织中,充分考虑了效率,挖掘算法的设计也考虑了实时性的要求,在挖掘结果的展示中借助COGNOS的强大功能,为前端员工提供了高效、实时的界面,以期对企业的信息化建设探索一条有本地特色的可行之路。本论文的主要工作有:(1)设计一个符合生产条件的数据仓库,实现对元数据的管理。借助ORACLE的新特性,以进程流实现对数据的逐步归约。为提高挖掘精度和实现界面响应高效做基础准备。(2)根据中国电信业务特点和用户消费习惯,以回归模式对用户数据进行挖掘,解决传统算法中通过提高测试集来提高挖掘精度和离线分析的方法,满足实际生产的要求。(3)在展示时,将COGNOS集成在.NET的框架中,实现业务智能和使用者的个性需求。更多还原

【Abstract】 After the reorganization of the domestic telecommunications industry, the market environment has become more reasonable and more intense competition between telecom providers in China, Customers are telecom providers survival and development foundation, retain customers and compete has become the focus of competition.As a full service provider, China Telecom’s business development in the C network, leading to some decline of tone business in cable telephone, and because the traditional business is post-paid way, C network users not to accept prepaid most of them to enterprise arrears has brought great pressure on the risk and income growth.To take into account customer perception and profits, to achieve differentiation and precision services for customers, the application of cable network universal reminder of the arrears management, the introduction of methods of banking credit, and improving brand image and target management of arrears .This dissertation studies the content of the data warehouse is based on the daily consumption of user behavior analysis and mining, were intrinsic regularity to realize the difference between customer service and management.In the telecommunications industry, widespread has a kind of phenomenon of "information rich", "knowledge poor", the traditional approach difficult to find it contains the knowledge of the urgent need for new knowledge, new technology to solve the problem. Information with data warehouse management, and based on the information to mining, which applies to client management, sales strategy, resource allocation and other aspects. In the real production environment, efficient access to information and business intelligence is the mining results can be applied to promote the crucial factor.The dissertation, the data warehouse design and organization, fully taking into account the efficiency of mining algorithm is also designed to consider the real-time requirements, the mining results demonstrate the power of using COGNOS for the marketing employees provides a highly efficient real-time interface ,and to the enterprise’s information construction to explore with local characteristics of a practical way.The main work of this thesis:(1) Designed data warehouse production conditions, to achieve the management of metadata. With ORACLE new features, in order to process the data stream to achieve a gradual reduction. To improve mining accuracy and achieve interface to prepare to respond efficiently.(2) According to China Telecom’s business characteristics and user habits, a kidn of online regression algorithm be used to improve depends testing set to improve the mining accuracy of the traditional and off-line analysis method algorithm, to meet the production of Practical requirements.(3) In the web, COGNOS integrated in the. NET framework, to achieve business intelligence and individual needs of users更多还原