數(shù)據(jù)挖掘應(yīng)用課件

數(shù)據(jù)挖掘應(yīng)用CRM顧客生命周期壽命盈利

獲取消費(fèi)者保持消費(fèi)者消費(fèi)者分析和恢復(fù)收入支出壽命CustomeridentificationCRMbeginswithcustomeridentification.Thisphaseinvolvestargetingthepopulationwhoaremostlikelytobecomecustomersormostprofitabletothecompany.Italsoinvolvesanalyzingcustomerswhoarebeinglosttothecompetitionandhowtheycanbewonback.Elementsforcustomeridentificationincludetargetcustomeranalysisandcustomersegmentation.CustomerattractionOrganizationscandirecteffortandresourcesintoattractingthetargetcustomersegments.Directmarketingisapromotionprocesswhichmotivatescustomerstoplaceordersthroughvariouschannels.directmailorcoupon目標(biāo)營(yíng)銷客戶流失分析CustomerdevelopmentElementsofcustomerdevelopmentincludecustomerlifetimevalueanalysis,up/crosssellingandmarketbasketanalysis.Customerlifetimevalueanalysisisdefinedasthepredictionofthetotalnetincomeacompanycanexpectfromacustomer.Up/Crosssellingreferstopromotionactivitieswhichaimataugmentingthenumberofassociatedorcloselyrelatedservicesthatacustomeruseswithinafirm.Marketbasketanalysisaimsatmaximizingthecustomertransactionintensityandvaluebyrevealingregularitiesinthepurchasebehaviourofcustomers.Personalizedrecommendationsystems

Informationfilteringandrecommendationrule-basedfiltering,content-basedfiltering,andcollaborativefiltering.Rule-basedfilteringusespre-specifiedif-thenrulestoselectrelevantinformationforrecommendation.Content-basedfilteringuseskeywordsorotherproduct-relatedattributestomakerecommendations.Collaborativefilteringusespreferencesofsimilarusersinthesamereferencegroupasabasisforrecommendation.TypicalpersonalizationprocessunderstandingcustomersthroughprodeliveringpersonalizedofferingbasedontheknowledgeabouttheproductandthecustomermeasuringpersonalizationimpactInadequateinformationinIROnepossiblesolutionforovercomingtheproblemistoexpandthequerybyaddingmoresemanticinformationtobetterdescribetheconcepts.Relevancefeedbacksandknowledgestructureareusedtoaddappropriatetermstoexpandthequeries.Relevancefeedbacksareinformationontheitemsselectedbytheuserfromtheoutputofpreviousqueries.Apersonalizedknowledgerecommendationsystem

Asemantic-expansionapproachtobuildtheuserproanalyzingdocumentspreviouslyreadbytheperson.Thesemantic-expansionapproachthatintegratessemanticinformationforspreadingexpansionandcontent-basedfilteringfordocumentrecommendation.Asamplesemantic-expansionnetworkExperimentalresultsAnempiricalstudyusingmasterthesesintheNationalCentrallibraryinTaiwanshowsthatthesemantic-expansionapproachoutperformsthetraditionalkeywordapproachincatchinguserinterests.自適應(yīng)構(gòu)件檢索構(gòu)件檢索是構(gòu)件庫(kù)研究中的重要問題，有效的構(gòu)件檢索機(jī)制能夠降低構(gòu)件復(fù)用成本。構(gòu)件的復(fù)用者并不是構(gòu)件的設(shè)計(jì)者或構(gòu)件庫(kù)的管理員，在檢索構(gòu)件時(shí)對(duì)構(gòu)件庫(kù)的描述理解不充分，導(dǎo)致難以給出完整和精確的檢索需求。用戶選擇構(gòu)件的結(jié)果反映其真實(shí)需求，如果能夠從用戶的檢索行為以及用戶對(duì)檢索結(jié)果的反饋中推斷出用戶的非精確檢索條件與用戶實(shí)際需要的精確檢索條件之間內(nèi)在聯(lián)系的模式，就可以提高系統(tǒng)的查準(zhǔn)率。基于關(guān)聯(lián)挖掘的自適應(yīng)構(gòu)件檢索把關(guān)聯(lián)規(guī)則挖掘方法引入構(gòu)件檢索，從用戶檢索行為以及反饋中挖掘出非精確檢索條件與精確檢索結(jié)果之間的關(guān)聯(lián)規(guī)則，從而調(diào)整檢索機(jī)制，提高構(gòu)件檢索的查準(zhǔn)率。實(shí)例{windows}{windows,SQLServer}{Linux}{Linux,Mysql}{金融}{金融,SQLServer}{windows,金融}{windows,金融,SQLServer}零部件供應(yīng)商選擇如何選擇供應(yīng)商不僅決定了產(chǎn)品的質(zhì)量和成本,也決定了產(chǎn)品的銷售價(jià)格、維護(hù)費(fèi)用和用戶滿意程度。選擇供應(yīng)商一般以滿足時(shí)間約束的條件下最小化物流成本為目標(biāo)，沒有考慮零部件故障率與不同地域環(huán)境之間的相關(guān)性?；陉P(guān)聯(lián)規(guī)則的零部件供應(yīng)商選擇使用關(guān)聯(lián)規(guī)則挖掘算法，從產(chǎn)品維修記錄中，尋找不同供應(yīng)商提供的產(chǎn)品零部件及其組合在不同地域的頻繁故障模式。在生成供應(yīng)商選擇和配送方案過程中，利用這些頻繁故障模式，選擇合適的零部件供應(yīng)商組合，達(dá)到物流成本與產(chǎn)品維護(hù)成本的聯(lián)合優(yōu)化。采用決策樹挖掘出人員選拔規(guī)則CHAIDDecisiontreeforpredictingjobperformanceImprovingeducationImprovingteachingandlearningInstructorscanhavetroubleidentifyingtheirrealdifficultiesinlearning.Basedonthestudents’testingrecords,thesystemworkstoidentifyandfindthoseproblems,andthencomesupwithitssuggestionsfordesigningnewteachingstrategies.Assistteacherstoidentifystudents’specificdifficultiesandweaknessesinlearning.Helpsthestudenttofindouthisorherweakpointsinlearningandoffersimprovementrecommendations.ESLrecommenderteachingandlearningRight/wronganswerstatisticaltableForeverystudent,thesystemcreatesaright/wronganswerstatisticaltable:awronganswerisrepresentedby1andarightanswerby0.Summarytableofstudents’wronganswersTheright/wronganswerstatisticaltablesforrespectivestudentsareintegratedinasummarytableofstudents’wronganswers,andthesumvaluesinthetablearethenrankedindescendingordersoastoshowthedescendingdegreesofweaknessesthestudentshavecollectively.HierarchicalclusteringHierarchicalclusteringalgorithmisthenappliedtodatacollectedtosegmentthestudentsintoacertainnumberofclusters,orcategories,eachofwhichincludesstudentssharingthesameorsimilarcharacteristics.Allstudents’right/wronganswerstatisticaltablesClusteringanalysisAclusteringanalysisismadeofthedatainAllstudents’right/wronganswerstatisticaltables.Itisevidentthatthestudentswhosenumbersareenclosedinthefollowingseparateparenthesesbelongtodifferentclustersrespectively:(9,15,6,17,13,19,14,5);(22,23,4,3,21,11,24,20,7,1);(12,18,2,8,25,10,16).搜索引擎優(yōu)化搜索引擎優(yōu)化Theyareusuallynotsearchenginesbythemselves.Theclusteringengineusesoneormoretraditionalsearchenginestogatheranumberofresults;then,itdoesaformofpost-processingontheseresultsinordertoclusterthemintomeaningfulgroups.Thepost-processingstepanalyzessnippets,i.e.,shortdocumentabstractsreturnedbythesearchengine,usuallycontainingwordsaroundquerytermoccurrences.研討題閱讀后面參考文獻(xiàn)，分析案例使用的數(shù)據(jù)挖掘方法以及解決的主要問題。結(jié)合自己的實(shí)踐，說(shuō)明所在崗位對(duì)商務(wù)智能的需求（針對(duì)軟件工程碩士）。典型參考文獻(xiàn)(1)Chen-FuChien,Li-FeiChen.Dataminingtoimprovepersonnelselectionandenhancehumancapital:acasestudyinhigh-technologyindustry.ExpertSystemswithApplication,2008,(34):280-290Cristo′balRomero,Sebastia′nVentura,EnriqueGarc?′a.Dataminingincoursemanagementsystems:Moodlecasestudyandtutorial.Computers&Education51(2008)368–384Yang,C.C.etal.,Improvingschedulingofemergencyphysiciansusingdatamininganalysis,ExpertSystemswithApplications(2008),doi:10.1016/j.eswa.2008.02.069JangHeeLee,SangChanPark.Intelligentprofitablecustomerssegmentationsystembasedonbusinessintelligencetools.ExpertSystemswithApplications29(2005):145–152Chih-MingChen,Ying-LingHsieh,Shih-HsunHsu.Mininglearnerproassociationruleforweb-basedlearningdiagnosis.ExpertSystemswithApplications33(2007)6–22Bong-HorngVhu,Ming-ShianTsai,Cheng-SeenHo.Towardahybriddataminingmodelforcusterretention.Knowledge-BasedSystems20(2007)703–718DanielaGrigoria,FabioCasatib,MaluCastellanos,etal.Businessprocessintelligence.ComputersinIndustry53(2004)321–343DursunDelen,ChristieFuller,CharlesMcCann.Analysisofhealthcarecoverage:Adataminingapproach.Delen,D.etal.,Analysisofhealthcarecoverage:Adataminingapproach,ExpertSystems