資源描述：
《面向商務(wù)智能的數(shù)據(jù)挖掘及其應(yīng)用分析》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在工程資料-天天文庫。
1、AbstractBigdataerahasarrivedandmassdataischallengingthetraditionalenterprisemanagementmodeandmarketingmode.Intheageofinformation,inordertobetterusehugedatainformationandmakewisedecision,enterprisesshouldeffectivelydealwiththedataandconvertitintovaluablebusinessinformation
2、.Inthisway,businessintelligenceisborn.Itmeanstohelpenterprisemakebetterbusinessdecisionthroughutilizeenterprisersdataproperty.AnapplicationresearchofdatamininginthefieldofbusinessintelligenceisbasedonmarketsegmentationinCRM.Throughtheprocessofextraction,loadinghugeinforma
3、tionrelatedtoconsumerbehavior,analystpredictdemandsforcustomersindifferentsegmentsandthisisbasedonindividualpurchasingbehaviorsandcustomersegmentation.Thispaperisbasedonthemodelingofmarketinganalysisandcustomerrelationship.Somedataminingmodelcontaincustomersurvivalanalysi
4、smodel,customerloyaltymodel,hierarchicalmodelaswellasanalysisandcomparisonofinfluenceofdifferentsalespromotionmode.Thisarticlefirstlydescribeshowpromotionalmodeinfluencesdailysalesbasedontwodifferentpromotionalmodes.Thearticleusesmultiplelinearregression,fixedeffects,Step
5、wiseregressionanalysistodeterminesignificanteffectvariable.Besides,itanalyzesthepurchasingbehaviorofConAgraandPopSecretfordifferentpromotionalmodelsleadtodifferentcustomerpurchasingbehaviors.Theresearchfindsthatfrequentpromotioncanhelpcustomersrecallbrands.Notallnewcustom
6、ersintroducedthroughpromotionactivitiesarepricesensitivecustomersandwefindthatintheirfollowingpurchases,theyalsotrytobuysomenon-promotionalproducts.Inthepartofclusteringanalysis,thispapertriedtwomethodstodividethecustomersintodifferentcategoriesandsummarizesthefeaturesofc
7、ustomersindifferentcategories.Intheprocessofclassification,weusedafastclusteringmethodandWardfsmethodtosegmentcustomersandresultsareexpressedbyscatterdiagram.ComparedtoWardt，fastclusteringmethodismoresuitableforthecustomercategory.Asquareshapedscatterplotshowsthatcategori
8、esareverynea匚Withineachcategory,theobservationsarenotsoconcentrated.SoIchooseafastclusteringmeth
