the applications and its strategies of adaboost algorith

the applications and its strategies of adaboost algorith

ID:5796049

大小:289.50 KB

页数:9页

时间:2017-12-25

the applications and its strategies of adaboost algorith_第1页
the applications and its strategies of adaboost algorith_第2页
the applications and its strategies of adaboost algorith_第3页
the applications and its strategies of adaboost algorith_第4页
the applications and its strategies of adaboost algorith_第5页
资源描述:

《the applications and its strategies of adaboost algorith》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、TheApplicationsanditsStrategiesofAdaBoostAlgorithmintheCreditRatingsonFeatureSelection:TakingIronandSteelSectorForExampleLIHui1,21SchoolofStatistics,RenminUniversityofChina,Beijing,PRC,100872;2SchoolofScienceandInformation,QingdaoAgriculturalUniversity,Q

2、ingdao,PRC,266109Abstract:Inthispaper,Adaboostalgorithmisappliedincreditratingsfirstly,andempiricalanalysisshowsthatAdaboostalgorithmonthebasisof18indexesselectedasregressionvariablesfitsthecreditratesof39listedironandsteelcompaniesofChinaverywell.Thedis

3、criminationerrorsare2.56%after10iterations,giveniterationsaddedthenitserrorscouldreachzeroandoutputclassificationresultsstably.Inadditions,theindeximportanceoutputsusedfromtwoaspectscanreselectandrefinetheninekeyindexesamongtheeighteenindexes.AfterusingA

4、daboostalgorithmtotestagain,wecanfindthatthenineindexesreduceddonotcutdowntheclassificationinformationofthemodelsandtheratingcorrectionsdonotslidedowneither.Keywords:creditratings;featureselection;AdaBoostalgorithms;ironandsteelsectorⅠ.IntroductionMostof

5、regressionmodels,whichreflectrealisticeconomyandsocietyevents,arerelatetofeatureselection,andthencreditratinginvolvesthem.Theessenceofcreditratingsistoprovideariskysymboltoinvestors,discoveringanddeliveringsomeinformationofcreditrisks.Therefore,creditrat

6、ingdescribescreditriskbycreditsymbols,whichcouldbeconsideredasakindofclassifications,andthatisakindofqualitativeregressions.Fromfeatureselectiononqualitativeregression,wecanfindqualitativeregressionisboundupwithfeatureselectionandgoodorbadofselectionwill

7、influentscorrectionandeffectsdirectly.Howevermodernfeatureselectionmethodsaredividedasthreeclasses:Filter,Wrapper,Embedded[1].Filterisakindofgeneralfeatureselectionmethod,anditusessomescoresonindexesimportancetoaccomplishdatapreprocessing.BeforeFilterisu

8、sedtomodel,ascorestatisticshouldbebuiltthenrankedtoallindexes.Atlast,someunimportantindexeswillbecancelledunderstandardpredetermined.SomecommonFilterincludesPearsoncorrelationcoefficient,T-test,chi-squaretestandInformation

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。