




版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
机器学习基础聚类基础()2020/12/3集群第8-1课集群基础定义和动机()数据预处理和相似度计算聚类目标聚类评价2020/12/3集群第8-2课集群基础定义和动机寻找一组对象,使得一组中的对象彼此相似(或相关),而与其他组中的对象不同(或无关)2020/12/3集群第8-3课集群基础定义和动机一个独立的工具:探索数据分发其它算法的预处理步骤模式识别,空间数据分析,图像处理,市场研究,万维网,……群集文档群集web日志数据以发现相似访问模式的组聚类共表达基因市场营销:帮助营销人员在他们的客户群中发现截然不同的群体,然后利用这些知识来制定有针对性的营销方案气候:了解地球气候,发现大气和海洋的模式2020/12/3集群第8-4课集群基础定义和动机一个独立的工具:探索数据分发其它算法的预处理步骤模式识别,空间数据分析,图像处理,市场研究,万维网,…两个重要方面输入数据的属性定义点之间的相似或相异聚类要求确定目标和方法2020/12/3集群第8-5课集群基础定义和动机数据预处理与相似性计算()聚类目标聚类评价2020/12/3集群第8-6课数据预处理和相似度计算数据:数据对象及其属性的集合属性是一个对象的属性或特征例如:一个人的眼睛颜色,温度等属性也称为维度,变量,字段,特性或特征属性的集合描述一个对象对象也称为记录,点,事例,样本,实体或实例2020/12/3集群第8-7课数据预处理和相似度计算数据矩阵()表示n个具有p个变量的对象2020/12/3集群第8-8课数据预处理和相似度计算相似与相异相似性两个数据对象相似程度的数值度量当物体更相似时,会更高通常落在[0,1]范围内相异两个数据对象差异的数值度量物体更相似时更低最小相异度通常为0上限各不相同2020/12/3集群第8-9课数据预处理与相似度计算DistanceMatrix(距离矩阵)Representspairwisedistanceinnobjects
Annbynmatrixd(i,j):
distanceordissimilaritybetweenobjectsiandjNonnegativeCloseto0:similar
2023/11/4ClusteringLesson8-10DataPreprocessingandSimilarityComputationDataMatrix->DistanceMatrix2023/11/4ClusteringLesson8-11DataPreprocessingandSimilarityComputationTypesofAttributes(属性的类型)Discrete(离散)HasonlyafiniteorcountablyinfinitesetofvaluesExamples:zipcodes,counts,orthesetofwordsinacollectionofdocumentsNote:binaryattributesareaspecialcaseofdiscreteattributesOrdinal(定序)HasonlyafiniteorcountablyinfinitesetofvaluesOrderofvaluesisimportantExamples:rankings(e.g.,painlevel1-10),grades(A,B,C,D)Continuous(连续)HasrealnumbersasattributevaluesExamples:temperature,height,orweightContinuousattributesaretypicallyrepresentedasfloating-pointvariables2023/11/4ClusteringLesson8-12DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributes
2023/11/4ClusteringLesson8-13DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributesMinkowskiDistance—ContinuousAttributeMinkowskidistance:ageneralizationIfq=2,disEuclideandistance欧几里德距离Ifq=1,disManhattandistance曼哈顿距离2023/11/4ClusteringLesson8-14DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributes
MinkowskiDistance—ContinuousAttributeStandardizationCalculatethemeanabsolutedeviationCalculatethestandardizedmeasurement(z-score)2023/11/4ClusteringLesson8-15DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributesMinkowskiDistance—ContinuousAttributeStandardizationMahalanobisDistance马氏距离具有相同分布的两个随机向量x和y与协方差矩阵S的相异测度。
Adissimilaritymeasurebetweentwo
randomvectorsxandyofthesame
distributionwiththe
covariancematrix
S.2023/11/4ClusteringLesson8-16DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributesMinkowskiDistance—ContinuousAttributeStandardizationMahalanobisDistanceCommonPropertiesofaDistanceDistances,suchastheEuclideandistance,havesomewellknownproperties1.d(p,q)>=0forallpandqandd(p,q)=0onlyifp=q.(Positivedefiniteness)2.d(p,q)=d(q,p)forallpandq.(Symmetry)3.d(p,r)<=d(p,q)+d(q,r)forallpointsp,q,andr.(TriangleInequality)2023/11/4ClusteringLesson8-17DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributesMinkowskiDistance—ContinuousAttributeMahalanobisDistanceSimilarityforBinaryAttributesComputesimilaritiesusingthefollowingquantitiesM01=thenumberofattributeswherepwas0andqwas1M10=thenumberofattributeswherepwas1andqwas0M00=thenumberofattributeswherepwas0andqwas0M11=thenumberofattributeswherepwas1andqwas1SimpleMatchingandJaccardCoefficients2023/11/4ClusteringLesson8-18DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributesMinkowskiDistance—ContinuousAttributeMahalanobisDistanceSimilarityforBinaryAttributesCommonPropertiesofaSimilaritys(p,q)=1(ormaximumsimilarity)onlyifp=q.s(p,q)=s(q,p)forallpandq.(Symmetry)wheres(p,q)is
thesimilaritybetweenpoints(dataobjects),pandq.2023/11/4ClusteringLesson8-19DataPreprocessingandSimilarityComputationSimilarity/DissimilarityforSimpleAttributesMinkowskiDistance—ContinuousAttributeMahalanobisDistanceSimilarityforBinaryAttributesCommonPropertiesofaSimilarityCharacteristicsoftheInputDataAreImportantSparseness,Attributetype,TypeofData,Dimensionality,NoiseandOutliers,TypeofDistribution=>Conductpreprocessingandselecttheappropriatedissimilarityorsimilaritymeasure进行预处理,选择合适的相似性或相似性度量=>Determinetheobjectiveofclusteringandchoosetheappropriatemethod确定聚类目标,选择合适的聚类方法2023/11/4ClusteringLesson8-20ClusteringBasicsDefinitionandMotivationDataPreprocessingandSimilarityComputationObjectiveofClustering(聚类目标)ClusteringEvaluation2023/11/4ClusteringLesson8-21ObjectiveofClusteringConsiderationsforClusterAnalysisPartitioningcriteriaSinglelevelvs.hierarchicalpartitioning(often,multi-levelhierarchicalpartitioningisdesirable)SeparationofclustersExclusive(e.g.,onecustomerbelongstoonlyoneregion)vs.overlapping(e.g.,onedocumentmaybelongtomorethanonetopic)HardversusfuzzyInfuzzyclustering,apointbelongstoeveryclusterwithsomeweightbetween0and1Weightsmustsumto1Probabilisticclusteringhassimilarcharacteristics
2023/11/4ClusteringLesson8-22ObjectiveofClusteringConsiderationsforClusterAnalysisRequirementsofClusteringScalabilityAbilitytodealwithdifferenttypesofattributesMinimalrequirementsfordomainknowledgetodetermineinputparametersAbletodealwithnoiseandoutliersDiscoveryofclusterswitharbitraryshapeInsensitivetoorderofinputrecordsHighdimensionalityIncorporationofuser-specifiedconstraintsInterpretabilityandusability
2023/11/4ClusteringLesson8-23ObjectiveofClusteringConsiderationsforClusterAnalysisRequirementsofClusteringNotionofaClustercanbeAmbiguous
2023/11/4ClusteringLesson8-24ObjectiveofClusteringConsiderationsforClusterAnalysisRequirementsofClusteringNotionofaClustercanbeAmbiguousTypesofClusters基于中心簇是一组对象,这样簇中的一个对象比任何其他簇的中心更接近(更相似)簇的“中心”星团的中心通常是一个质心,即星团中所有点的平均值,或者是一个星团中最具代表性的点2023/11/4ClusteringLesson8-25ObjectiveofClusteringConsiderationsforClusterAnalysisRequirementsofClusteringNotionofaClustercanbeAmbiguousTypesofClustersCenter-based
Density-based基于密度Aclusterisadenseregionofpoints,whichisseparatedbylow-densityregions,fromotherregionsofhighdensity.Usedwhentheclustersareirregularorintertwined,andwhennoiseandoutliersarepresent.2023/11/4ClusteringLesson8-26ClusteringBasicsDefinitionandMotivationDataPreprocessingandSimilarityComputationObjectiveofClusteringClusteringEvaluation(聚类评价)2023/11/4ClusteringLesson8-27ClusteringEvaluationClustervalidationQuality:“goodness”ofclustersAssessthequalityandreliabilityofclusteringresultsWhyvalidation?ToavoidfindingclustersformedbychanceTocompareclusteringalgorithmsTochooseclusteringparameterse.g.,thenumberofclusters2023/11/4ClusteringLesson8-28ClusteringEvaluationAspectsofClusterValidation
Comparingtheclusteringresultstogroundtruth(externallyknownresults)–ExternalIndex(外部指标)Evaluatingthequalityofclusterswithoutreferencetoexternalinformation–Useonlythedata–InternalIndex(内部指标)Determiningthereliabilityofclusters–Towhatconfidencelevel,theclustersarenotformedbychance–Statisticalframework2023/11/4ClusteringLesson8-29ClusteringEvaluationComparingtoGroundTruth(与真值比较)NotationN:numberofobjectsinthedatasetP={P1,…,Ps}:thesetof“groundtruth”clustersC={C1,…,Ct}:thesetofclustersreportedbyaclusteringalgorithmThe“incidencematrix”(关联矩阵)NbyN(bothrowsandcolumnscorrespondtoobjects)Pij=1
ifOiandOjbelongtothesame“groundtruth”clusterinP;Pij=0otherwise
Cij=1ifOiandOjbelongtothesameclusterinC;Cij=0otherwise2023/11/4ClusteringLesson8-30ClusteringEvaluationComparingtoGroundTruthNotationThe“incidencematrix”(关联矩阵)RandIndexandJaccardCoefficientApairofdataobject(Oi,Oj)fallsintooneofthefollowingcategoriesSS:Cij=1andPij=1;(agree)DD:Cij=0andPij=0;(agree)SD:Cij=1andPij=0;(disagree)DS:Cij=0andPij=1;(disagree)2023/11/4ClusteringLesson8-31ClusteringEvaluationComparingtoGroundTruthNotationThe“incidencematrix”(关联矩阵)RandIndexandJaccardCoefficientEntropyandPuritythenumberofobjectsinboththek-thclusteroftheclusteringsolutionandj-thclusterofthegroundtruththenumberofobjectsinthek-thclusteroftheclusteringsolutionthenumberofobjectsinthej-thclusterofthegroundtruth2023/11/4ClusteringLesson8-32ClusteringEvaluationComparingtoGroundTruthInternalIndex(内部指标)UseonlythedatatomeasureclusterqualityMeasurethe“cohesion”and“separation”ofclustersCalculatethecorrelationbetweenclusteringresultsanddistancematrix2023/11/4ClusteringLesson8-33ClusteringEvaluationComparingtoGroundTruthInternalIndexCohesionandSeparationCohesionismeasuredbythewithinclustersumofsquaresSeparationismeasuredbythebetweenclustersumofsquares2023/11/4ClusteringLesson8-34ClusteringEvaluationComparingtoGroundTruthInternalIndexCohesionandSeparationCohesionismeasuredbythewithinclustersumofsquaresSeparationismeasuredbythebetweenclustersumofsquaresBSS+WSS=constantWSS(Cohesion)measureiscalledSumofSquaredError(SSE)—acommonlyusedmeasureAlargernumberofclusterstendtoresultinsmallerSSE2023/11/4ClusteringLesson8-35ClusteringEvaluationComparingtoGroundTruthInternalIndexCohesionandSeparationSilhouetteCoefficient(轮廓系数)SilhouetteCoefficientcombinesideasofbothcohesionandseparation.Foranindividualpoint,iCalculatea=averagedistanceofitothepointsinitsclusterCalculateb=min(averagedistanceofitopointsinanothercluster)Thesilhouettecoefficientforapointisthengivenbys=1–a/bifa<b,(s=b/a-1ifa>b,nottheusualcase)Typicallybetween0and1Thecloserto1thebetterCancalculatetheAverageSilhouettewidthforaclusteroraclustering2023/11/4ClusteringLesson8-36ClusteringEvaluationComparingtoGroundTruthInternalIndex(内部指标)CohesionandSeparation
SilhouetteCoefficient(轮廓系数)CorrelationwithDistanceMatrixDistanceMatrixDijisthesimilaritybetweenobjectOiandOjIncidenceMatrixCij=1ifOiandOjbelongtothesamecluster,Cij=0otherwise
ComputethecorrelationbetweenthetwomatricesOnlyn(n-1)/2entriesneedstobecalculatedHighcorrelationindicatesgoodclustering2023/11/4ClusteringLesson8-37ClusteringEvaluationComparingtoGroundTruthInternalIndexCohesionandSeparation
SilhouetteCoefficient(轮廓系数)CorrelationwithDistanceMatrixGivenDistanceMatrixD={d11,d12,…,dnn}andIncidenceMatrixC={c11,c12,…,cnn}.CorrelationrbetweenDandCisgivenby2023/11/4ClusteringLesson8-38ClusteringEvaluationComparingtoGroundTruthInternalIndexCohesionandSeparation
SilhouetteCoefficient(轮廓系数)CorrelationwithDistanceMatrixUsingSimilarityMatrixforClusterValidation
Orderthesimilaritymatrixwithrespecttoclusterlabelsandinspectvisually.2023/11/4ClusteringLesson8-39ClusteringEvaluationComparingtoGroundTruthInternalIndexReliabilityofClustersNeedaframeworktointerpretanymeasure–Forexample,ifourmeasureofevaluationhasthevalue,10,isthatgood,fair,orpoor?StatisticsprovideaframeworkforclustervalidityThemore“atypical”aclusteringresultis,themorelikelyitrepresentsvalidstructureinthedata
2023/11/4ClusteringLesson8-40ClusteringEvaluationComparingtoGroundTruthInternalIndexReliabilityofClustersStatisticalFrameworkforSSEExampleCompareSSEof0.005againstthreeclustersinrandomdataSSEHistogramof500setsofrandomdatapointsofsize100—lowestSSEis0.01732023/11/4ClusteringLesson8-41ClusteringEvaluationComparingtoGroundTruthInternalIndexReliabilityofClustersStatisticalFrameworkforSSEDeterminetheNumberofClustersUsingSSE2023/11/4ClusteringLesson8-42sklearn.metrics.pairwiseThesklearn.metrics.pairwisesubmoduleimplementsutilitiesto
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 材料专业试题及答案
- 锅炉专业试题及答案
- 汽修专业大专试题及答案
- 编辑类专业试题及答案
- 广东省2025-2026学年高三上学期9月月考历史试题(含答案)
- 编导专业笔试题及答案
- 广东省江门市广雅中学2024-2025学年七年级上学期10月份英语月考卷(含答案无听力原文及音频)
- 化妆品行业品牌发展战略
- 全国重点城市领导讲话稿模板
- 短视频行业内容创作与营销策略
- 深度解析Palantir介绍
- 小学五爱教育
- 《探寻抗日战争历史》课件
- 玻璃幕墙维修保养施工方案
- 开学第一课消防安全知识课件
- 亲子关系断绝协议书范文模板
- 香港劳务派遣合同范本年
- 2025年威海桃威铁路有限公司招聘笔试参考题库含答案解析
- 妊娠期高血压疾病的观察及护理
- 临床骨筋膜室综合征护理业务学习
- 医院DIP支付方式改革工作实施方案
评论
0/150
提交评论