PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文

PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文

ID:39593245

大小:2.61 MB

页数:22页

时间:2019-07-06

PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文_第1页
PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文_第2页
PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文_第3页
PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文_第4页
PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文_第5页
资源描述:

《PREDICTABILITY_ RECENT INSIGHTS FROM INFORMATION THEORY英文》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、PREDICTABILITY:RECENTINSIGHTSFROMINFORMATIONTHEORY1,23TimothyDelSoleandMichaelK.TippettReceived25April2006;revised27December2006;accepted17April2007;published21November2007.[1]Thispapersummarizesaframeworkforinvestigatinganalysisbetweenforecastandobservations.T

2、heregressionpredictabilitybasedoninformationtheory.Thisframeworkoperatorthatmapsobservationstoforecastsplaysanconnectsandunifiesawidevarietyofstatisticalmethodsimportantroleinthisframework,withtheleftsingulartraditionallyusedinpredictabilityanalysis,includingli

3、nearvectorsofthisoperatorbeingthepredictablecomponentsregression,canonicalcorrelationanalysis,singularvalueandthesingularvaluesbeingthecanonicalcorrelations.decomposition,discriminantanalysis,anddataassimilation.Thiscorrespondencebetweenpredictablecomponentsand

4、Centraltothisframeworkisaprocedurecalledpredictablesingularvectorsoccursonlyifthesingularvectorsarecomponentanalysis(PrCA).PrCAoptimallydecomposescomputedusingMahalanobisnorms,aresultthatshedsvariablesbypredictability,justasprincipalcomponentlightontheroleofnor

5、msinpredictability.Inlinearanalysisoptimallydecomposesvariablesbyvariance.Forstochasticmodelstheforcingthatminimizespredictabilitynormaldistributionsthesamepredictablecomponentsareistheonethatrendersthe‘‘whitened’’dynamicaloperatorobtainedwhetheroneoptimizespre

6、dictiveinformation,thenormal.Thisconditionforminimumpredictabilityisdispersionpartofrelativeentropy,mutualinformation,invarianttolineartransformationandisequivalenttoMahalanobiserror,averagesignaltonoiseratio,normalizeddetailedbalance.Theframeworkalsoinspiresso

7、menewmeansquareerror,oranomalycorrelation.Forjointnormalapproachestoaccountingfordeficienciesofforecastmodelsdistributions,PrCAisequivalenttocanonicalcorrelationandestimatingdistributionsfromfinitesamples.Citation:DelSole,T.,andM.K.Tippett(2007),Predictability:

8、Recentinsightsfrominformationtheory,Rev.Geophys.,45,RG4002,doi:10.1029/2006RG000202.1.INTRODUCTION2002].Despitebeingsmall,predictablevariancebeyond3weeksmaystillbeofinterest

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。