實驗報告一多元線性回歸分析采用青海省海北牧業(yè)氣象試驗站3月18日至10月28日23旬的土壤濕度、旬降水、旬平均氣溫的資料，用SAS對青海省海北地區(qū)土壤濕度與旬降水、旬平均氣溫進行多元線性回歸分析。原始數(shù)據(jù)如下：青海省海北地區(qū)土壤濕度與旬平均降水、氣溫的關(guān)系09:14Saturday,June12,20041Obsyraintemp12414.51722658.716330920.91942326.161520521.1111622734.197728133.650822538.0106919126.11161021236.41241122013.21281222212.21311321855.51401429565.81481529747.91461626939.5131172259.51171826123.8951927149.8942024863.383212093.7662223126.737232362.35y:土壤濕度；rain:旬降水;temp:旬平均氣溫解答：編寫程序：datashidu;inputyraintemp@@;cards;2414.5172658.71630920.9192326.16120521.111122734.19728133.65022538.010619126.111621236.412422013.212822212.213121855.514029565.814829747.914626939.51312259.511726123.89527149.89424863.3832093.76623126.7372362.35 ;PROCREG；Modely=raintemp；Run；輸出結(jié)果：SAS系統(tǒng)2009年04月16日星期四下午09時27分56秒1TheREGProcedureModel:MODEL1DependentVariable:v1NumberofObservationsRead23NumberofObservationsUsed23AnalysisofVarianceSumofMeanSourceDFSquaresSquareFValuePr>FModel26647.216563323.608284.120.0318Error2016148807.38700CorrectedTotal2222795RootMSE28.41456R-Square0.2916DependentMean243.04348AdjR-Sq0.2208CoeffVar11.69114ParameterEstimatesParameterStandardVariableDFEstimateErrortValuePr>|t|Intercept1244.9378113.4598218.20<.0001v211.015820.370252.740.0125v31-0.341720.15681-2.180.0414

結(jié)果分析：方差分析表中，Sr=6647.21656,Se=16148,自由度為2和20，F(xiàn)=3323.60828/807.38700=4.12,且服從自由度（2，20）的F分布隨機變量大于4.12的概率為0.0318<0.05,所以回歸是顯著的。在參數(shù)估計中，截距、降水、氣溫三者的T檢驗都達到0.05的顯著水平，所以回歸系數(shù)是顯著的。因此得出回歸方程為：y=244.94+1.02*rain-0.34*temp2．從過去23年的氣象資料查得表3.7．試找出主要的預(yù)報因子(自變量)和相應(yīng)回歸方程。表3.7x1x2x3x4x5x6y317165426523305474262233310000258325460626220266125726026279194926627277194525919311342225763181022661628140042652225161847268243012524262302452298264282831924262243000002642427214482593026107392621730111022609286705260202992215259163213001263920712052511634760325716解答：編寫程序：datarain;inputx1-x6y;cards;317165426523305474262233310000258325460626220266125726026279194926627277194525919311342225763181022661628140042652225161847268243012524262302452298264282831924262243000002642427214482593026107392621730111022609286705260202992215259163213001263920712052511634760325716;procreg;modely=x1-x6/selection=stepwise;run;輸出結(jié)果：SAS系統(tǒng)2009年04月16日星期四下午09時27分56秒3TheREGProcedureModel:MODEL1DependentVariable:yNumberofObservationsRead23NumberofObservationsUsed23StepwiseSelection:Step1Variablex5Entered:R-Square=0.3751andC(p)=10.5463AnalysisofVarianceSumofSourceDFSquaresSquareFValuePr>FModel1455.21540455.2154012.600.0019Error21758.5237336.12018CorrectedTotal221213.73913ParameterStandardVariableEstimateErrorTypeIISSFValuePr>Intercept11.747632.51246789.6807421.860.0x51.709660.48159455.2154012.600.0019Boundsonconditionnumber:1,1----------------------------------------------------------------------------------------------------------------------------------StepwiseSelection:Step2Variablex2Entered:R-Square=0.4960andC(p)=6.8267AnalysisofVarianceSumofMeanSourceDFSquaresSquareFValuePr>FModel2602.04987301.024949.840.0011Error20611.6892630.58446CorrectedTotal221213.73913ParameterStandaVariableEstimateErrorTypeIISSFValuePr>FIntercept17.278373.42294779.3074825.48<.0001x2-0.644560.29417146.834474.800.04x51.626890.44476409.2296513.380.0016Boundsonconditionnumber:1.0073,4.0291----------------------------------------------------------------------------------------------------------------------------------StepwiseSelection:Step3Variablex6Entered:R-Square=0.6488andC(p)=1.6018

SAS系統(tǒng)2009年04月16日星期四下午09時27分56秒4TheREGProcedureModel:MODEL1DependentVariable:yStepwiseSelection:Step3AnalysisofVarianceSumofMeSourceDFSquaresSquareFValuePr>FModel3787.53155262.5105211.700.0001Error19426.2075822.43198CorrectedTotal221213.73913ParameterStandardVariableEstimateErrorTypeIISSFValuePr>FIntercept-188.0846471.47778155.321716.920.0164x2-0.772910.25585204.712209.130.0070x51.443280.38621313.2695313.970.0014x60.792890.27574185.481678.270.0097Boundsonconditionnumber:1.0548,9.3879----------------------------------------------------------------------------------------------------------------------------------Allvariablesleftinthemodelaresignificantatthe0.1500level.Noothervariablemetthe0.1500significancelevelforentryintothemodel.SummaryofStepwise