07FPAdvanced.pptVIP免费

11DataMining:ConceptsandTechniques(3rded.)—Chapter7—JiaweiHan,MichelineKamber,andJianPeiUniversityofIllinoisatUrbana-Champaign&SimonFraserUniversity©2011Han,Kamber&Pei.Allrightsreserved.January26,2025DataMining:ConceptsandTechniques23Chapter7:AdvancedFrequentPatternMiningPatternMining:ARoadMapPatternMininginMulti-Level,Multi-DimensionalSpaceConstraint-BasedFrequentPatternMiningMiningHigh-DimensionalDataandColossalPatternsMiningCompressedorApproximatePatternsPatternExplorationandApplicationSummaryResearchonPatternMining:ARoadMap45Chapter7:AdvancedFrequentPatternMiningPatternMining:ARoadMapPatternMininginMulti-Level,Multi-DimensionalSpaceMiningMulti-LevelAssociationMiningMulti-DimensionalAssociationMiningQuantitativeAssociationRulesMiningRarePatternsandNegativePatternsConstraint-BasedFrequentPatternMiningMiningHigh-DimensionalDataandColossalPatternsMiningCompressedorApproximatePatternsPatternExplorationandApplicationSummary6MiningMultiple-LevelAssociationRulesItemsoftenformhierarchiesFlexiblesupportsettingsItemsatthelowerlevelareexpectedtohavelowersupportExplorationofsharedmulti-levelmining(Agrawal&Srikant@VLB’95,Han&Fu@VLDB’95)uniformsupportMilk[support=10%]2%Milk[support=6%]SkimMilk[support=4%]Level1min_sup=5%Level2min_sup=5%Level1min_sup=5%Level2min_sup=3%reducedsupport7Multi-levelAssociation:FlexibleSupportandRedundancyfilteringFlexiblemin-supportthresholds:SomeitemsaremorevaluablebutlessfrequentUsenon-uniform,group-basedmin-supportE.g.,{diamond,watch,camera}:0.05%;{bread,milk}:5%;…RedundancyFiltering:Somerulesmayberedundantdueto“ancestor”relationshipsbetweenitemsmilkwheatbread[support=8%,confidence=70%]2%milkwheatbread[support=2%,confidence=72%]ThefirstruleisanancestorofthesecondruleAruleisredundantifitssupportisclosetothe“expected”value,basedontherule’sancestor8Chapter7:AdvancedFrequentPatternMiningPatternMining:ARoadMapPatternMininginMulti-Level,Multi-DimensionalSpaceMiningMulti-L...