数据挖掘在超市客户关系管理中的应用

华中科技大学硕士学位论文数据挖掘在超市客户关系管理中的应用姓名：连欢申请学位级别：硕士专业：系统工程指导教师：陈学广20080606IWTOMicrosoftMicrosoftB/SMicrosoftMicrosoftIIAbstractAsthedevelopmentofinformationtechnology,therelationshipbetweenenterpriseandInformationSystemisbecomingmoreandmoretightness.Furthermore,sinceChinatokepartinWTO,manymultinationalcorporationshavejoinedinthemarketofChina.Inthatcase,theenterprisesofChinafacemuchgreatercompetition,whichbringsthemhighstressaswellaschance.Astherearelargeamountofoperationdata,andtheforeigncompanieshavebeguntousebusinessintelligencewhichfindsusefulbusinessinformationandruleshiddenindatabase,thetraditionalDecisionSupportSystem(DSS)can’tfeedtherequiresofmanagersofChinaenterprise.CombinationofDataWarehouse(DW),On-LineAnalyticalProcessing(OLAP)andDataMining(DM)isoneofthenewdirectionsofDDS,andDWisprincipalpartofthesetechnologies.InthisnewDSS,DWstoresdataandmanagestheresourcesbythemeansofdatabase,drawstheinformationwithOLAP,discoverstheknowledgeandrulesbythearithmeticofDM.ToestablishDWisthepreconditionofscientificdecision,isalsocrucialweapontogetsuperiorityincompetition.First,thisthesisintroducesthebackgroundknowledgeofOLAPandDMaswellasDSS,andintroducesMicrosoftDecisionTreearithmeticandMicrosoftClusterarithmeticintheDMpart.BasedonapplyingthelimitationsoftraditionalDSS,itputsforwardasetofnewsolutionthatcombineDSSOLAPandDMtoformatypeofnewDSS.Afterthat,itdesignsanddevelopstheB/S(Browser/Server)DSSofSupermarketwiththeACCESSdatabaseofafoodsupermarket,anddescribestheprocessofbuildingDWandCubeindetails.Besidesthefunctionofmanagementofbasedinformation,itbuildstheCustomerClassifymodelbasedonMicrosoftClusterarithmeticandtheMemberCardmodelbasedonMicrosoftDecisionarithmetic,whicharethefoundationsofDMimplementationtotheDSS.Asthereishugeamountofdatainsupermarket,itisagoodplatformfortheimplementationofDM.ThisthesisputsforwardafeasibleimplementationtotheCRMClientRelationshipManagementofsupermarket,andhastheactualreferencevalueforthepromotionofDMbasedonSQLServer2005.Keywords:DecisionSupportSystem,ClientRelationshipManagement,On-LineAnalyticalProcessing,DataMining_____111.1(DecisionSupportSystems,DSS)[1]DSS(On-LineTransactionProcessing,OLTP)[2]DSS(DataWarehouse,DW)[3](DataMining,DM)(On-LineAnalyticalProcessing,OLAP)[4]RalphKimball:[1],2[5]1.23[5]:(DW)+(OLAP)(DM)[6][6]B/S1.32070(ManagementInformationSystem,MIS)[7]DSSDSSDSSDSS4DSS[8]DSSMicrosoftSQLServer2005AnalysisServicesOLAP2080DSS[8][9](DW)(0LAP)(DM)(CRM)ERP[9](1)(2)(3)(4)(5)(6)(7)OLAPGUI()/DSS[1]ERP()CRM()SCM()BI()[10]CRM(ClientRelationshipManagement)[11]5GartnerGroup1999CRMCRM[11]CRM[][12]1.41.4.11MicrosoftSQLServer2005AnalysisServices62,MicrosoftSQLServer2005BusinessIntelligenceDevelopmentStudio3B/S(Browser/Server,/)Web1.4.2SQLServer2005MicrosoftOLAPSQLServer2005WebWebB/S()72OLAPOLAPSQLServer20052.12.1.12070M.S.ScottMortonDSSDSSKeenScottMorton1978DSS:DSS[8]DSSDSS[5]82.1.2[1]2.1:2.1(1):(2):(3):(4):DSSWeb(5):2.1.3DSSOLAPDSS[1,13]:1.DSS92.DSSDSS3.DSSSQLDSS4DSSDSS2.22090[14]OLAPDSSDSSDSS--:102.2DSSOLAPOLAPDSSDSSDSS[15,16,17]OLAP[18,19]112.3OLAPEF.Codd1993[20]Codd(OLTP)SQUCoddOLAPOLAPOnlineAnalyticalProcessing[21]Decision-SupportSystemDSSOLAPOLAPSalesCustomerProductTimeStoreCustomerCountryStateCityUnitSalesStoreSalesProfitCostOLAPOLAPOLAP2.3.1[2122]CustomerEducation12EducationEducation_idDescriptionCustomerEducationEducation_idEducationEducationCustomerEducation_id2.3.2[2123]Sale_Fact3Store_SalesUnit_SalesStore_CostSumMinMaxAverageDistinctCount[24]sliceSQLServer2005SQLServer2005salesinventory2.3.3[21]13Store_SalesStore_SalesAnalysisServicesSQL2005AnalysisServices3OLAPMultidimensionalOLAPMOLAPAnalysisServicesMOLAPMOLAPOLAPRelationalOLAPROLAPRDBMSROLAPMOLAPROLAPMOLAPOLAPHybridOLAPHOLAPHOLAPRDBMSMOLAPAnalysisServicesAnalysisServicesAnalysisServicesSQLRDBMS[2122]2.3.4OLAP14MOLAPAnalysisServices2005proactivecachingMOLAPROLAP[21]OLAPROLAPMOLAPMOLAPMOLAPROLAPMOLAPMOLAPMOLAPMOLAPMOLAPMOLAP2.3.5OLAPOLEDBOLAPMultiDimensionalExpressionMDXMDXSelect…From…WhereMDXSQLMDXSQLSQLSQLMDXMDXSelectWhereMDX[21]152.42.4.1[21](DM:DataMining)[25][26]()[2728][29]2.4.2[21]151002noise163liftchart4567GartnerResearch817SQLServer20052.32.4.3[30][21](TreeBuilding)[31](TreePruning)18;:()[32]1ID3[33]ID3()m12m2i1IS,S,...,SPlog(P)ii==−∑(2-1)mmiCi=12mAv{1ava}ASvijSjSiCAv1j2jmj1j2jmjj=1SS...SE(A)=-I(S,S,...,S)S++∑(2-2)191j2jmjSS...SS++jAjaSm1j2jmjij2iji=1I(S,S,...,S)=-Plog(P)∑2-3ijijjSP|S|=jSiCA12mGain(A)=I(S,S,...,S)-E(A)2-4ID3S2C4.5ID3C4.5[33]K:1Gain(A)GainRatio(A)SplitI(A)=vj2jj1SplitI(A)Plog(P)==−∑2-52C4.5A12ma,a,...a(m)An20jv(j=1,2,,n)jvjv1.C4.52.ifthen3MicrosoftMicrosoft[21]MicrosoftMicrosoft10099MicrosoftComplexity_Penalty01MicrosoftMicrosoftnn99212.1MAXIMUM_INPUT_ATTRIBUTES255MAXIMUM_OUTPUT_ATTRIBUTES255SCORE_METHOD3SPLIT_MECHOD3MINIMUM_SUPPORT10COMPLEXITY_PENALTYFORCED_REGRESSOR2.4.4(clustering)(class)(cluster)[21]221MicrosoftK-meanexpectationmaximization,EMK-meansKhardclustering,EMK-meansEM23Data05101520253035400510152025303540Clusters0510152025303505101520253035402.42.5225%2.212345%70%35%0%15%5%45%30%50%25%0%25%30%0%2