搜索
《金融数据挖掘案例分析》课程设计报告学生姓名：学号：题目：基于分类技术的电信客户流失预测的研究系别：信息管理与工程系班级：信息管理与信息系统指导教师：2015年6月信息管理与工程系课程设计1目录摘要：....................................................................................................................1Abstract：.............................................................................................................21.引言....................................................................................................................31.1研究背景和意义..........................................................................................31.2国内外应用和研究现状.................................................................................31.3研究方法..................................................................................................41.3.1数据收集..........................................................................................41.3.2数据准备..........................................................................................41.3.3数据建模..........................................................................................41.3.4.模型评估..........................................................................................42.决策树算法基本概述.............................................................................................52.1决策树算法的提出和发展...........................................................................52.2决策树算法的概念...................................................................................52.3决策树的优缺点..........................................................................................53.数据预处理........................................................................................................63.1数据预处理概述..........................................................................................63.2数据的商业理解........................................................................................63.3数据预处理方法........................................................................................63.3.1数据清理..........................................................................................63.3.2数据集成..........................................................................................73.3.3数据变换..........................................................................................73.3.4数据归约..........................................................................................74.构造决策树..........................................................................................................84.1决策树分类的步骤.......................................................................................84.2建模...........................................................................................................84.2.1输入数据..........................................................................................84.2.2输出类型..........................................................................................9信息管理与工程系课程设计24.2.3手工计算验证....................................................................................94.2.4SQLServerBusinessIntelligenceDevelopmentStudio工具验证...124.2.5实验结论分析..................................................................................165总结与后需改进工作............................................................................................175.1总结.........................................................................................................175.2后续需要改进的工作..................................................................................17致谢.....................................................................................................................19参考文献...............................................................................................................20信息管理与工程系课程设计11基于分类技术的电信客户流失预测的研究摘要：在国内随着对数据挖掘技术的重视，数据挖掘技术的应用也越来越广，其中电信行业的客户流失分析就更是一大热点。通过对以往流失客户的数据进行分析，找出可能流失用户的特征，及时采取相应的措施，减少客户流失的发生。这对提高经营业绩和降低运营成本有着极为重要的价值。本文从数据挖掘的效率和精度出发，运用分类技术方法中的决策树算法对电信客户的属性特征进行分析，得出流失客户的基本特征，以帮助企业管理者对该类客户的行为特征进行分析，采取针对性的措施挽留即将流失的客户或有流失意向的客户，达到亡羊补牢的效果。关键字：数据挖掘；电信客户流失；分类技术；决策树算法信息管理与工程系课程设计2Abstract：Alongwithdataminingtechnologydevelopment,dataminingimpor-tancealreadybymoreandmoremanypersonattention，inwhichtothetelecommunicationprofessioncustomeroutflowforecastisapresentbighotspot.Thisarticleutilizesthedecisiontreealgorithmtocarryontheanalysistothetelecommunicationcustomerattributecharacteristic,obtainstheoutflowcustomerthebasiccharacteristic,helpstheenterprisesuperintendenttocarryontheanalysistothiskindofcustomerbehaviorcharacteristic,adoptsthecustomerwhichthepointedmeasuredetainssoondrainsorhastheoutflowintentioncustomer,achievedisbetterlatethannevereffect.Keywords:Datamining;telecommunicationcustomeroutflow;classificationtechnique；decisiontreealgorithm信息管理与工程系课程设计31.引言1.1研究背景和意义随着中国电信行业体制的改革与重组，中国电信业的市场环境发生了根本性的变化，中国电信服务市场逐步形成了从最初个别运营商垄断市场到数家大运营商主导、多家小运营商参与、新运营商不断加入的电信市场竞争的新格局。在当前电信业普及率很高的形势下，在发展新客户的同时，怎样维持已有的客户群，已经成为电信企业越来越关注的焦点。面对激烈的竞争市场，各大运营商正在寻找一种更有效的办法来建立与客户的关系，创造客户价值来保留和竞争优质客户。要想预测将要流失的客户，进而成功对其进行挽留，首先要全面掌握客户的信息。这些业务数据已经达到几十甚至上百TB，数据挖掘技术则正是目前数