資源描述：
《2 frequent pattern -a.ppt》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、WhatIsFrequentPatternAnalysis?Frequentpattern:apattern(asetofitems,subsequences,substructures,etc.)thatoccursfrequentlyinadatasetFirstproposedbyAgrawal,Imielinski,andSwami[AIS93]inthecontextoffrequentitemsetsandassociationruleminingMotivation:Findinginher
2、entregularitiesindataWhatproductswereoftenpurchasedtogether?—Beeranddiapers?!WhatarethesubsequentpurchasesafterbuyingaPC?WhatkindsofDNAaresensitivetothisnewdrug?Canweautomaticallyclassifywebdocuments?ApplicationsBasketdataanalysis,cross-marketing,catalogd
3、esign,salecampaignanalysis,Weblog(clickstream)analysis,andDNAsequenceanalysis.WhyIsFreq.PatternMiningImportant?DisclosesanintrinsicandimportantpropertyofdatasetsFormsthefoundationformanyessentialdataminingtasksAssociation,correlation,andcausalityanalysisS
4、equential,structural(e.g.,sub-graph)patternsPatternanalysisinspatiotemporal,multimedia,time-series,andstreamdataClassification:associativeclassificationClusteranalysis:frequentpattern-basedclusteringDatawarehousing:icebergcubeandcube-gradientSemanticdatac
5、ompression:fasciclesBroadapplicationsAMultidimensionalViewofFrequentPattenDiscoverytypesofdataorknowledgelatticetransversal/mainoperationsothersassociativepatternsequentialpatternicebergcubereadwritepointotherinterestmeasurecompressionmethodpruningmethodc
6、onstraintsclosed/maxpatternSub-Graphpattern關(guān)聯(lián)規(guī)則基本概念A(yù)priori及其改進算法Apriori-basedSub-GraphMining本講內(nèi)容DataandKnowledgeTypesAssociativePatterntransactionaltablevsrelationaltablebooleanvsquantitativeSequentialPatternAsequence:<(ef)(ab)(df)cb>(e,f)->(a,b)->coccur5
7、0%ofthetimeIceburgCubetablewithameasuresTIDItems10a,c,d20b,c,e30a,b,c,e40b,eTIDabcde1010110200010130111014001001transactional=binaryMonthCityCust_grpProdCostPriceJanTorEduPrinter500485MarVanEduHD540520………………relationalwithquantitativeattributeMonthCityCust
8、_grpProdCost(Support)JanTor*Printer1040……………cube:usingothermeasureassupportSimulationofLatticeTransversala,ca,eb,eb,ca,bc,eb,c,ea,b,ea,c,ea,b,cabcea,b,c,e{}Thewholeprocessoffrequentpatternminingcanbeseenasasearchint
