Classification and regression trees

Classification and regression trees

ID:40600386

大小:474.99 KB

页数:10页

时间:2019-08-04

Classification and regression trees_第1页
Classification and regression trees_第2页
Classification and regression trees_第3页
Classification and regression trees_第4页
Classification and regression trees_第5页
资源描述:

《Classification and regression trees》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、OverviewClassificationandregressiontreesWei-YinLohClassificationandregressiontreesaremachine-learningmethodsforconstructingpredictionmodelsfromdata.Themodelsareobtainedbyrecursivelypartitioningthedataspaceandfittingasimplepredictionmodelwithineachpartition.Asaresult,thep

2、artitioningcanberepresentedgraphicallyasadecisiontree.Clas-sificationtreesaredesignedfordependentvariablesthattakeafinitenumberofunorderedvalues,withpredictionerrormeasuredintermsofmisclassifica-tioncost.Regressiontreesarefordependentvariablesthattakecontinuousororderedd

3、iscretevalues,withpredictionerrortypicallymeasuredbythesquareddifferencebetweentheobservedandpredictedvalues.Thisarticlegivesanin-troductiontothesubjectbyreviewingsomewidelyavailablealgorithmsandcomparingtheircapabilities,strengths,andweaknessintwoexamples.C2011JohnW

4、iley&Sons,Inc.WIREsDataMiningKnowlDiscov2011114–23DOI:10.1002/widm.8CLASSIFICATIONTREESXtakesorderedvalues,thesetSisanintervaloftheform(−∞,c].Otherwise,Sisasubsetofthevaluesnaclassificationproblem,wehaveatrainingsam-IpleofnobservationsonaclassvariableYthattakenbyX.Thep

5、rocessisappliedrecursivelyonthedataineachchildnode.Splittingstopsiftherelativetakesvalues1,2,...,k,andppredictorvariables,decreaseinimpurityisbelowaprespecifiedthreshold.X1,...,Xp.Ourgoalistofindamodelforpredict-Algorithm1givesthepseudocodeforthebasicsteps.ingthevalueso

6、fYfromnewXvalues.Intheory,thesolutionissimplyapartitionoftheXspaceintokAlgorithm1Pseudocodefortreeconstructiondisjointsets,A1,A2,...,Ak,suchthatthepredictedbyexhaustivesearchvalueofYisjifXbelongstoAj,forj=1,2,...,k.IftheXvariablestakeorderedvalues,twoclassical1.Starta

7、ttherootnode.solutionsarelineardiscriminantanalysis1andnear-22.ForeachX,findthesetSthatminimizesestneighborclassification.ThesemethodsyieldsetsthesumofthenodeimpuritiesinthetwoAjwithpiecewiselinearandnonlinear,respectively,childnodesandchoosethesplit{X∗∈S∗}boundariestha

8、tarenoteasytointerpretifpislarge.thatgivestheminimumoverallXandS.ClassificationtreemethodsyieldrectangularsetsAjbyrecursively

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。