资源描述:
《数据挖掘关联规则》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、Chapter4:MiningFrequentPatterns,AssociationandCorrelationsBasicconceptsandaroadmapScalablefrequentitemsetminingmethodsMiningvariouskindsofassociationrulesConstraint-basedassociationminingFromassociationtocorrelationanalysisMiningcolossalpatternsSummary2021/7/191DataMining:ConceptsandTechniquesWhat
2、IsFrequentPatternAnalysis?Frequentpattern:apattern(asetofitems,subsequences,substructures,etc.)thatoccursfrequentlyinadatasetFirstproposedbyAgrawal,Imielinski,andSwami[AIS93]inthecontextoffrequentitemsetsandassociationruleminingMotivation:FindinginherentregularitiesindataWhatproductswereoftenpurch
3、asedtogether?—Beeranddiapers?!WhatarethesubsequentpurchasesafterbuyingaPC?WhatkindsofDNAaresensitivetothisnewdrug?Canweautomaticallyclassifywebdocuments?ApplicationsBasketdataanalysis,cross-marketing,catalogdesign,salecampaignanalysis,Weblog(clickstream)analysis,andDNAsequenceanalysis.2021/7/192Da
4、taMining:ConceptsandTechniques关联规则挖掘关联规则挖掘的典型案例:购物篮问题在商场中拥有大量的商品(项目),如:牛奶、面包等,客户将所购买的商品放入到自己的购物篮中。通过发现顾客放入购物篮中的不同商品之间的联系,分析顾客的购买习惯哪些物品经常被顾客购买?同一次购买中,哪些商品经常会被一起购买?一般用户的购买过程中是否存在一定的购买时间序列?具体应用:利润最大化商品货架设计:更加适合客户的购物路径货存安排:实现超市的零库存管理用户分类:提供个性化的服务2021/7/193DataMining:ConceptsandTechniques关联规则挖掘简单的说,关联规
5、则挖掘就是发现大量数据中项集之间有趣的关联在交易数据、关系数据或其他信息载体中,查找存在于项目集合或对象集合之间的频繁模式、关联、相关性、或因果结构。应用购物篮分析、交叉销售、产品目录设计、聚集、分类等两种策略:1。商品放近,增加销量2。商品放远,增加其他商品的销量2021/7/194DataMining:ConceptsandTechniquesWhyIsFreq.PatternMiningImportant?Freq.pattern:AnintrinsicandimportantpropertyofdatasetsFoundationformanyessentialdatamining
6、tasksAssociation,correlation,andcausalityanalysisSequential,structural(e.g.,sub-graph)patternsPatternanalysisinspatiotemporal,multimedia,time-series,andstreamdataClassification:discriminative,frequentpatternanalysisClusteranalysis:frequentpattern-basedclusteringDatawarehousing:icebergcubeandcube-g
7、radientSemanticdatacompression:fasciclesBroadapplications2021/7/195DataMining:ConceptsandTechniques关联规则挖掘形式化定义给定:?设I={i1,i2,…,im}是项(item)的集合。若干项的集合,称为项集(ItemSets)?记D为交易(transaction)T(或事务)的集合,这里交易T是项的集合,并且T⊆I。对应每一