I 摘要 客户是银行的生存之本，是实现竞争力提升和可持续发展的基础。在互联网金 融高速发展、第三方支付抢占客户资源、同业竞争力巨大、银行 到店客户率越来越 低的新形势下，银行的经营模式必须发生改变，必须由过去的B2C模式转变为C2B 模式，即由过去的银行设计产品或服务推销给客户，转变为通过深入挖掘分析 银行 经过几十年经营发展沉淀的庞大历史客户数据，找到客户精准需求，为客户设计或 匹配产品或服务。 本文以客户关系管理理论为基础，结合Y银行X支行目前客户关系管理的 现状 和问题，提出了改进的AFPR细分模型。随机筛选Y银行X支行658位客户自2018 年5月1日至2019年4月30日共计99743条交易数据进行实证研究，采用熵值法 确定模型中四个变 量的权重，并结合K-means聚类算法对样本客群进行细分，将客 户分为五类。结合支行工作实际，分析聚类客户特征，将五类客户定义为高价值类 客群、活期结算类客群、理财类 客群、易流失类客群、社区类客群，最后针对上述 五类客群分别采取不同的客户关系管理策略。 本文完善了商业银行客户关系管理体系，为今后商业银行通过数据挖掘建立模 型 ，找出相似客户群体并实现精准营销提供了一定的理论基础和实践意义。 关键词客户；RFM模型；数据挖掘；熵值法；K-means聚类 河北科技大学硕士学位论文 II Abstract Customeristhebasisofthebank'ssurvival,anditisthebasisforachieving competitivenessandsustainabledevelopment.WiththerapiddevelopmentofInternet finance,third-partypaymentspreemptingcustomerresources,thehugecompetitivenessof theindustry,andtheincreasinglylowbank-to-storecustomerrate,thebusinessmodelof banksmustbechangedandchangedfromthepreviousB2CmodeltotheC2Bmodel. Thatis,fromthepastbankdesignproductsorservicessoldtocustomers,toin-depth miningandanalysisofthebank'shugehistoricalcustomerdataafterdecadesof developmentanddevelopment,tofindthepreciseneedsofcustomersanddesignormatch productsorservicesforcustomers. Basedonthetheoryofcustomerrelationshipmanagementandthecurrentsituation andproblemsofcustomerrelationshipmanagementinXbranchofYbank,thispaper proposesanimprovedAFPRsubdivisionmodel.Randomscreeningof658customersof theXbranchoftheYBankfromMay1,2018toApril30,2019,atotalof99743 transactiondatawereexperimentallystudied.Theentropymethodwasusedtodetermine theweightsofthefourvariablesinthemodel,andtheK-meansclusteringalgorithmwas combined.Thegroupofsamplecustomersissubdividedandthecustomersaredivided intofivecategories.Accordingtothepracticeofbranchwork,thecharacteristicsofcluster customersareanalyzed,andthefivetypesofcustomersaredefinedashigh-valueguest groups,currentsettlementguestgroups,financialmanagementguestgroups,easilylost guestgroups,andcommunityguestgroups.Finally,differentcustomerrelationship managementstrategiesareadoptedfortheabovefivetypesofguestgroups. Thispaperimprovesthecustomerrelationshipmanagementsystemofcommercial banksandprovidesatheoreticalbasisandpracticalsignificanceforcommercialbanksto establishamodelthroughdatamining,findoutsimilarcustomergroupsandrealize accuratemarketing. KeywordsCustomer；RFMmode；Datamining;Entropymethod；K-meansclustering 目录 III 目录 第1章绪 论...................................................................................................................1 1.1研究背景和研究意 义..............................................................................................1 1.1.1研究背 景...........................................................................................................1 1.1.2研究意 义...........................................................................................................2 1.2研究内容及研究方 法..............................................................................................2 1.2.1研究内 容...........................................................................................................2 1.2.2研究方 法...........................................................................................................2 1.3.2创新之 处...........................................................................................................4 第2章理论基础和文献综 述...........................................................................................5 2.1客户关系管理理 论..................................................................................................5 2.2客户细分理 论..........................................................................................................5 2.3客户价值理 论..........................................................................................................6 2.4数据挖掘理 论..........................................................................................................6 2.5RFM模型理 论.........................................................................................................7 2.6熵值法理 论..............................................................................................................8 2.7K-means聚类算 法...................................................................................................9 2.8本章小 结................................................................................................................10 第3章Y银行X支行客户关系管理现状及问题分 析.................................................11 3.1Y银行X支行简 介................................................................................................11 3.1.1X支行基本情 况.............................................................................................11 3.1.2X支行经营概 况.............................................................................................11 3.2Y银行X支行客户关系管理现 状.......................................................................13 3.2.1客户结构现 状.................................................................................................13 3.2.2客户关系管理系统使用现 状.........................................................................15 3.3Y银行X支行客户关系管理存在的问 题...........................................................17 3.3.1以“客户为中心的理念”易说难 做.................................................................17 3.3.2综合营销意识薄弱，客户产品持有种类 少.................................................17 3.3.3客户信息维护不及时，关键信息完整率低.................................................18 3.3.4客户 分层体系简单粗暴，差别化服务效果不理想.....................................19 河北科技大学硕士学位论文 IV 3.4本章小 结................................................................................................................19 第4章Y银行X支行客户细分实证分 析....................................................................21 4.1创建客户细分模 型................................................................................................21 4.2样本数据分 析........................................................................................................21 4.2.1数据来源分 析.................................................................................................21 4.2.2数据统计分 析.................................................................................................25 4.3数据预处 理............................................................................................................27 4.3.1数据清 洗.........................................................................................................27 4.3.2属性选 择.........................................................................................................27 4.3.3数据标准 化.....................................................................................................28 4.4熵值法确定客户价 值............................................................................................30 4.5聚类分 析................................................................................................................31 4.5.1挖掘工具的选 择.............................................................................................31 4.5.2聚类结果分 析...........................................