返回
chap8_basic_cluster_analysis基本聚类分析课件.ppt

chap8_basic_cluster_analysis基本聚类分析课件.ppt

ID：57055698

大小：2.53 MB

页数：104页

时间：2020-07-30

chap8_basic_cluster_analysis基本聚类分析课件.ppt_第1页
chap8_basic_cluster_analysis基本聚类分析课件.ppt_第2页
chap8_basic_cluster_analysis基本聚类分析课件.ppt_第3页
chap8_basic_cluster_analysis基本聚类分析课件.ppt_第4页
chap8_basic_cluster_analysis基本聚类分析课件.ppt_第5页
资源描述：

《chap8_basic_cluster_analysis基本聚类分析课件.ppt》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、DataMining ClusterAnalysis:BasicConcepts andAlgorithmsLectureNotesforChapter8IntroductiontoDataMiningbyTan,Steinbach,Kumar©Tan,Steinbach,KumarIntroductiontoDataMining4/18/20041WhatisClusterAnalysis?Findinggroupsofobjectssuchthattheobjectsinagroupwillbesimila

2、r(orrelated)tooneanotheranddifferentfrom(orunrelatedto)theobjectsinothergroupsInter-clusterdistancesaremaximizedIntra-clusterdistancesareminimizedApplicationsofClusterAnalysisUnderstandingGrouprelateddocumentsforbrowsing,groupgenesandproteinsthathavesimilarf

3、unctionality,orgroupstockswithsimilarpricefluctuationsSummarizationReducethesizeoflargedatasetsClusteringprecipitationinAustraliaWhatisnotClusterAnalysis?SupervisedclassificationHaveclasslabelinformationSimplesegmentationDividingstudentsintodifferentregistra

4、tiongroupsalphabetically,bylastnameResultsofaqueryGroupingsarearesultofanexternalspecificationGraphpartitioningSomemutualrelevanceandsynergy,butareasarenotidenticalNotionofaClustercanbeAmbiguousHowmanyclusters?FourClustersTwoClustersSixClustersTypesofCluster

5、ingsAclusteringisasetofclustersImportantdistinctionbetweenhierarchicalandpartitionalsetsofclustersPartitionalClusteringAdivisiondataobjectsintonon-overlappingsubsets(clusters)suchthateachdataobjectisinexactlyonesubsetHierarchicalclusteringAsetofnestedcluster

6、sorganizedasahierarchicaltreePartitionalClusteringOriginalPointsAPartitionalClusteringHierarchicalClusteringTraditionalHierarchicalClusteringNon-traditionalHierarchicalClusteringNon-traditionalDendrogramTraditionalDendrogramOtherDistinctionsBetweenSetsofClus

7、tersExclusiveversusnon-exclusiveInnon-exclusiveclusterings,pointsmaybelongtomultipleclusters.Canrepresentmultipleclassesor‘border’pointsFuzzyversusnon-fuzzyInfuzzyclustering,apointbelongstoeveryclusterwithsomeweightbetween0and1Weightsmustsumto1Probabilisticc

8、lusteringhassimilarcharacteristicsPartialversuscompleteInsomecases,weonlywanttoclustersomeofthedataHeterogeneousversushomogeneousClusterofwidelydifferentsizes,shapes,anddensitiesTypesofClustersW

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。