一類復雜觀測數(shù)據(jù)模擬的統(tǒng)計診斷及對策.pdf

1、河海大學碩士學位論文一類復雜觀測數(shù)據(jù)模擬的統(tǒng)計診斷及對策姓名：石雯申請學位級別：碩士專業(yè)：應用數(shù)學指導教師：袁永生20070401AbstractInmanyareBsofseientitleresearchandactualproduction，thereexistsmuchworkaboutanalyzingandprocessingdataAnalyzingandprocessingcomplicatedobservatiolad

2、ataisveryvaluableandsignificanttoactualapplicationInthispaperaccordingtoakindofconcreteeomplicatedwaterandsedimentobservationdatawehascarriedOnthestatisticaldiagnosistothecomplicateddatawhiehstudiedestablishedthemulti1in

3、earregressionmodelandexaminedtheroodelatthesametimewediagnosisthismodelbasedonthepredecessorresearchworkBecauslethereisaseriousproblemofthemulticorrelationsbetweenindependentvailablestheordinaryleastsquaresmethodhasthepo

4、ssibilitytocausethemodeIunstabletopredictionConsideredthesampleeapaeityofthedataissmallTlloreoverthedatahasthemulticonelations，themodelofpartialleastsquaresregressionisestablishedusingthemethodofpartialleastsquaresregres

5、sionmeanwhile，consideredtheridgeregressioncaneffectivelyeliminatethemulticorrelations，themodelofridgeregressionisalSOestablishedComparingthetwokindofllewregressionroodelstotheordinaryleastsquaresmodeltheresultshowthatthe

6、methodofpartialleastsauaresregressionandridgeregressioncallreducethemeansquareofparametersestimationslrengthenmodelgeneralityandmakethesimulationeffecttobebetter。TIlisarticleluminescentspotisthatthemethodsofpartialleasts

7、quaresregressionandridgeregressioncallimprovethemodelthentheprecisionoftheseregressionmodelsiscomparedaccordingtothediagnosisofakindofcomplicatedobservationdataandinviewofthequestionelaaraeteristicTwokindofnewregressionm