多元统计分析实验报告计算协方差矩阵相关矩阵SAS.docx

（一）院 系：数学与统计学学院专 业：统计学年 级：2009级课程名称：统计分析学 号：姓 名：指导教师：2012年 4月28日（一）实验名称1.编程计算样本协方差矩阵和相关系数矩阵;根据多元分析结果，P指小于，表明在的显着水平下，四个变量有显 着差异。2.多元方差分析MANOVA。（二）实验目的1 .学习编制sas程序计算样本协方差矩阵和相关系数矩阵;2 .对数据进行多元方差分析。（三）实验数据第一题：xlx2x3x4x5x6x7446217818240621851854445156168424016617238551781804758176176407017618043641621704463174176384817018644451681684556186192455117617647471621645450166170494418018551571681725110481621684848162164497616816857581741765462156165524816416650481461555148172172544416817251591861885749148155495618618848521701765253170172第二题:kindxlx2x3x41125603382101119802333301635126020316551429150113065403205169453501901466058520011466627325018754585240111077507270110760364200113061391200180454292701605044219018154260280113587507260157484002851755252026017665403250155424111702665445531028245403210265653122802405147728026754481293238504682102424535119021134039031028055520200276605071892943326028026051429190255403902952654848117726948442225212563312270212056416280270454683702626641622426960377280365334802603100344682953656341626531174846825031146339538035530546235364515073203110904422253606244024831106937726038878299360373633903203114554942403103544163103100332733123140613123453803628625031355446834531306932536036057273260（四）实验内容1 .打开SAS软件并导入数据；2 .编制程序计算样本协方差矩阵和相关系数矩阵;3 .编制sas程序对数据进行多元方差分析；4 .根据实验结果解决问题，并撰写实验报告；（五）实验体会（结论、评价与建议等）第一题：程序如下：proc corr data= cov;proc corr data= nosimple cov;with x3 x4;partial xl x2;run;结果如下:(1)协方差矩阵7变量： xlx2x3x4x5x6x7协方差矩阵，自由度=30xlx2x3x4x5xGx7xlxl27.1591398-10.1364946-8.45626451.364709?-6.1193548-18.0516129-20.6752688x2x2-10.136494669.3650389-7.22105081.65825471.568204315.498655919.0337204x3x38.4562645-7.221050828.3793848-6.3725471-15.3064183-21.7352376-11.5574785x4 x41.36470971.6582547-6.37254711.92491784.60930114.46124732.8747634x5x5-6.11935481.5682043-15.30641834.609301168.797849527.038709719.5731183x6x6-18.051612915.4986559-21.73523764.461247327.0387097105.103225887.3505376x7x7-20.675268819.0337204-11.55747852.874763419.573118387.350537683.9806452SAS系统CORR过程08:15 Friday, April 28, 20001(2)相关系数矩阵Pearson相关系数,N = 31当 HO： Rho=0 时，Prob > |r|xlx2x3x4x5x6x7xl1.00000-0.23354-0.304590.18875-0.14157-0.33787-0.43292xl0.20610.09570.30920.44750.06300.0150x2-0.233541.00000-0.162750.143510.022700.181520.24938x20.20610.38170.44120.90350.32840.1781x3-0.30459-0.162751.00000-0.86219-0.34641-0.31797-0.23874x30.09570.3817<.00010.05630.02660.1997x40.188750.14351-0.862191.000000.400540.313650.22610x40.30920.4412<.00010.02560.08580.2213x5-0.141570.02270-0.346410.400541.000000.317970.25750x50.44750.90350.05630.02560.08130.1620xG-0.337870.18152-0.397970.313650.317971.000000.92975x60.06300.32840.02660.08580.0813<.0001x7-0.432920.24938-0.236740.226100.257500.929751.00000x70.01500.17610.19970.22130.1620<.0001第二题: 程序如下:proc anova data=;class kind;model xl-x4=kind;manova h=kind;run;结果如下：(1)分组水平信息The ANOVA ProcedureClass LeveI Informat ionCI assLeveIsVaiueskind31 2 3Number of observat ions 60(2) xl> x2、x3、x4的方差分析The ANOVA ProcedureDependent Variable： xl xlSourceModelErrorCorrected TotalSum ofDFSquares25221.300005744069.550005949290.85000Mean Square F Value2610.65000773.150003.38Pr > F0.0411xl Mean85.95000R-Square Coeff Var Root MSE0.10592832.3508727.80557R-Square Coeff Var0.05364222.99888Root MSE x2 Mean12.6685555.08333SourceDF Anova SS Mean Square F Valuekind2518.5333333259.26666671.62Pr > F0.2078SourcekindDF2Anova SS5221.300000Mean Square2610.650000F Value3.38Dependent Variable： x2 x2The ANOVA ProcedureSum ofSourceDFSquaresMean SquareF ValueModel2518.533333259.2666671.62ErrorCorrected Total57599148.0500009666.583333160.492105Pr > F0.0411Pr > F0.2078dependent Variable： x3 x3The ANOVA ProcedureSourceModelErrorCorrected TotalDF25759Sum of Squares2480.8333427028.5000429509.3333Mean Square1240.41677491.7281F Value0.17Pr > F0.8478R-SquareCoeff VarRootMSEx3Mean0.00577621.1798086.55477408.6667SourceDFAnova SSMean SquareF ValuePr > Fkind22480.8333331240.4166670.170.8478The ANOVA ProcedureDependent Variable： x4 x4Sum ofSourceDFSquaresMean SquareF ValuePr > FModel238529.300019264.65008.010.0009Error57137115.10002405.5281Corrected Total59175644.4000R-Square Coeff Var Root MSEx4 Mean0.21936018.9660449.04618258.6000SourcekindDF2Anova SS38529.30000Mean Square19264.65000F Value Pr > F8.010.0009（3）多元方差分析The ANOVA ProcedureMultivariate Analysis of VarianceCharacteristic Roots and Vectors of： E Inverse 米 H, whereH = Anova SSCP Matrix for kindE = Error SSCP MatrixCharacterist ic RootPercentCharacteristic Vector V，EV=1x3x4xlx20.3360453873.17-0.00045795-0.003790960.000303660.002793390.1232338326.830.004241110.002368780.000018420.000028320.000000000.000.00121062-0.000324010.00157046-0.000065390.000000000.00-0.003177980.010435260.000070140.00076972MANOVA Test Criteria and F Approximations for the Hypothesis of No OveralI kind EffectH = Anova SSCP Matrix for kindE = Error SSCP MatrixS=2M=0.5N=26Stat ist icValueF ValueNum DF Den DF Pr > FWilks' Lambda0.66635953Pillai's Trace0.36123585Hotel Iing-Law leyTrace0.45927921Roy's Greatest Root0.336045384372000633341080.00401100.004174.8560.0048550.0027NOTE： F Statistic for Roy s Greatest Root is an upperbound.NOTE： F Statistic for Wilks' Lambda is exact.