资源描述：
《大数据数据挖掘培训讲义3：概念,属性和实例.ppt》由会员上传分享，免费在线阅读，更多相关内容在PPT专区-天天文库。
1、Input: Concepts,Attributes, InstancesModuleOutlineTerminologyWhat’saconcept?Classification,association,clustering,numericpredictionWhat’sinanexample?Relations,flatfiles,recursionWhat’sinanattribute?Nominal,ordinal,interval,ratioPreparingtheinputARFF,attributes,missingvalues,gettingtoknowda
2、tawitten&eibe2TerminologyComponentsoftheinput:Concepts:kindsofthingsthatcanbelearnedAim:intelligibleandoperationalconceptdescriptionInstances:theindividual,independentexamplesofaconceptNote:morecomplicatedformsofinputarepossibleAttributes:measuringaspectsofaninstanceWewillfocusonnominaland
3、numericoneswitten&eibe3What’saconcept?DataMiningTasks(Stylesoflearning):Classificationlearning: predictingadiscreteclassAssociationlearning: detectingassociationsbetweenfeaturesClustering: groupingsimilarinstancesintoclustersNumericprediction: predictinganumericquantityConcept:thingtobelea
4、rnedConceptdescription:outputoflearningschemewitten&eibe4ClassificationlearningExampleproblems:attritionprediction,usingDNAdatafordiagnosis,weatherdatatopredictplay/notplayClassificationlearningissupervisedSchemeisbeingprovidedwithactualoutcomeOutcomeiscalledtheclassoftheexampleSuccesscanb
5、emeasuredonfreshdataforwhichclasslabelsareknown(testdata)Inpracticesuccessisoftenmeasuredsubjectively5AssociationlearningExamples:supermarketbasketanalysis-whatitemsareboughttogether(e.g.milk+cereal,chips+salsa)Canbeappliedifnoclassisspecifiedandanykindofstructureisconsidered“interesting”D
6、ifferencewithclassificationlearning:Canpredictanyattribute’svalue,notjusttheclass,andmorethanoneattribute’svalueatatimeHence:farmoreassociationrulesthanclassificationrulesThus:constraintsarenecessaryMinimumcoverageandminimumaccuracy6ClusteringExamples:customergroupingFindinggroupsofitemsth
7、ataresimilarClusteringisunsupervisedTheclassofanexampleisnotknownSuccessoftenmeasuredsubjectivelySepallengthSepalwidthPetallengthPetalwidthType15.13.51.40.2Irissetosa24.93.01.40.2Irissetosa…517.03.24.71.4Irisversicolor526.43.24.51.5Irisversicolor…1016.33.36.02
