線性混合模型的模型選擇.pdf

1、北京工業(yè)大學碩士學位論文線性混合模型的模型選擇姓名：王雨申請學位級別：碩士專業(yè)：概率論與數(shù)理統(tǒng)計指導教師：王松桂2003.5.1AbstractThereFessionmodelselectionisanimportaIltresearchdirectionofmodem印plicationstatisticItmainlyresearches恤erationalit)rofre蓼essionmodelandtheind印endentV

2、ariablesselectionLinearmixedmodelisa虹ndoflinearmodels，wllichincludesbothfixedefrectsandrandomefrcctsnis丘equentlyusedinbi010幽medicine，economics，detennining，sanlplingdesignsaIldqualitycontr01procedllres，aIldsoonPeopleusual

3、lypmallrelationalind印endentvariablesintotllereFessionmodelItmayincreasemeamountoftllere野essionmodelcalculation，a11dmaydecreasetheprecisionoft11eestimationandthepredictionThereforeitiscostlytoacquiresomeindependentVariabl

4、esdataandputthemintotllemodeltos01vesomepracticalprobleIIls，Thjsmesiswillfocusontllemodelselcctionpmbleminlinearmixedmodel，a11dputfonⅣardanewcriterionnlismesisusesmeMeaJlSquareEnDrMetllod，anditcons訊Ictst11eR“isedMSEcrit謝

5、onByrevje麗ngtlleRevisedMSE，wecanchoosetheind印endentvariablesofmere簪essionmodelaIlddecreasethecalculationAndthjscriteriondoesn’td印e11dondis研b砸onofmemodelhl也isthesis，mecri矧onis1lsedtos01ve也esalaryproblem，whjchisoneofthepro

6、blemsofH啪孤ResourceManagem饑tTheapplicationoftheRevisedMSEreceivesrightresults，a11dith嬲thes柵ee丘’ectwitllmeotllercrit鰣onncanbeusedasanewwayt0solvetllepolicyofcomp％sation趾dbenefitsyst鋤probleminHumanResOurceMallagementKeyword