Clustering-Presentation

Clustering-Presentation

ID:37578560

大小:1.95 MB

页数:244页

时间:2019-05-25

Clustering-Presentation_第1页
Clustering-Presentation_第2页
Clustering-Presentation_第3页
Clustering-Presentation_第4页
Clustering-Presentation_第5页
资源描述:

《Clustering-Presentation》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库

1、ClusterAnalysisAnalyzingMultivariateDataJLattin,JDCarroll,PEGreenIntroductionofClusterAnalysisClusteranalysisisanexploratoryprocedurethatisusedtoidentifygroupsofsimilarobjects(e.g.,people,stimuli,books,singers,etc.)inalargecollectionofobjects.Theidentifiedgroupshavemembersthataresimilartoeach

2、otheranddifferentfromthemembersinothergroups.TheapproachissimilarinspirittoMDS,butproducesdiscretegroupswithoutanyspatialrepresentation.Theidentifiedgroupscanbeusedinsubsequentanalyses.IntroductionofClusterAnalysisTheapproachisdecidedly“discovery”oriented—thereisusuallynopriorknowledgeofhowor

3、whyobjectsmightbedistributedintoparticulargroups.Themostcommonclusteringmethodsarehierarchicalandagglomerative—formingclustersbyjoiningnearbyobjectsorclusters,beginningwithasmanyclustersasthereareobjectsandendingwithasinglecluster.InbetweenthereMAYbeanumberofclustersthatprovidesaconvenientsim

4、plificationofthedata.IntroductionofClusterAnalysisTheclusteringsolutiondependson:•Methodofjoiningnearbyobjects•Typeofdistanceorsimilaritymeasureused(e.g.,Euclideandistance,squaredEuclideandistance,Minkowskimetric,correlation,binarymatches)•Theinformationcontainedinthedistanceorsimilaritymeasu

5、re•Natureofthedata(standardizedornon-standardized,bycaseorbyvariable)OutlineA.NONHIERARCHICALCLUSTERING:(k-mean)1).AlgorithmI.2).AlgorithmII.B.HIERARCHICALCLUSTERING1).Agglomerativehierarchicalclusteringalgorithms:1.Centroidmethod.2.Nearest-neighbororsingle-linkagemethod.3.Farthest-neighboror

6、complete-linkagemethod4.Average-linkagemethod5.Ward'smethod.2).Divisivehierarchicalclusteringalgorithms:1.monothetic.2.polythetic.Outline(continued)C.OTHERCLUSTERINGMETHODS1).SelfOrganizingMap(SOM).2).GraphTheoryBasedMethod3).......DefinitionClusteranalysisisatechniqueusedforcombiningobservat

7、ionsintogroupsorclusterssuchthat:that:•Eachgrouporclusterishomogeneous(orcompact)withrespecttocertaincharacteristics..•Eachgroupshouldbedifferentfromothergroupswithrespecttothesamecharacteristics..WhatIsGoodClustering?•Agoodclusteringmethodwi

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。
相关文章
更多
相关标签