实验报告实验项目名称典型相关分析所属课程名称统计分析及SAS实现实验类型验证性实验实验日期2016-12-11班级数学与应用数学学号姓名成绩实验概述:【实验目的及要求】掌握SAS中进行典型相关分析的方法,理解典型相关分析的基本思想。理解典型相关变量的性质。理解典型相关系数的求解步骤、特征根的求解,掌握典型相关系数的假设检验过程。掌握SAS系统中INSIGHT模块、“分析家”模块以及编程进行典型相关分析。【实验原理】SAS软件的操作方法及原理【实验环境】(使用的软件)SAS9.3实验内容:【实验方案设计】一.理解典型相关分析的概念及步骤;二.掌握典型相关分析的方法;三.用INSIGHT、“分析家”计算统计量和编程实现实际问题中的典型相关分析;【实验过程】(实验步骤、记录、数据、分析)【练习7-1】对某高中一年级男生38人进行体力测试及运动能力测试,如表所示,试对两组指标作典型相关分析。表体力测试、运动能力测试指标Numx1x2x3x4x5x6x7y1y2y3y4y514655126517525726.848927836025255954281.218507.246430334834669107389818746.8430329386449501054897.616606.836226633154255904666.52687.2453231139164861106437825587405297389749601004990.6156074202110379848631225256.117687.1466282362945551054876156174152463861048641203860.220627.14132873981149521004253.46427.44042364001247621003461.210627.24272574071341511015362.456083722534091452551254386.35626.84963010350154552945051.420657.63942433991649571104772.31945744630113371753651124790.415756.64463012357184777954772.39646.64202544471948601204786.412626.844728113812049551134184.1156073982743872148691284247.920637.14853073502242571224654.215637.24002863882354641555171.419616.951133122982453631204256.68537.54302943532542711384465.2175574872993702646661204562.222687.4470287360274556912966.218517.93802653582850601204256.68576.8460325348294251126505013577.33982723833048501154152.96397.44152863143142521404856.315606.950027113483248671053969.223607.645028103263349741514954.22058750030123303447551134071.419647.64102973313549741205354.522596.950033213423644521103754.914577.54002924213752661304745.914456.850528113553848681004553.623707.2522289352其中,体力测试指标为:X1-------反复横向跳(次),X2-------纵跳(cm),X3------背力(kg),X4------捏力(kg),X5-----台阶测试(指数),X6------定向体前屈(cm),X7-------俯卧上提后仰(cm)。运动能力测试的指标为y1-50m跑(s),y2-跳远(cm),y3-投球(m),y4引体向上(次),y5-耐力跑(s)。【解答】利用INSIGHT模块进行典型相关分析:结果:表7.1UnivariateStatisticsVariableNMeanStdDevMinimumMaximumy1387.13160.33546.60008.0000y238441.842143.2138362.0000522.0000y33827.81582.749521.000033.0000y4387.52633.83262.000021.0000y538366.605331.2976298.0000447.0000x13847.42113.366241.000054.0000x23860.21057.451350.000077.0000x338114.473715.659890.0000155.0000x43844.63165.354629.000053.0000x53866.910514.448545.900098.0000x63815.34215.95152.000025.0000x73859.73688.142939.000075.0000由表7.1得知一些基本统计量,各变量下的均值、标准差、最大值、最小值。表7.2CorrelationMatrixx1x2x3x4x5x6x7y1-0.3784-0.3791-0.3498-0.3184-0.4011-0.0719-0.2552y20.32380.52700.57890.2777-0.19460.25360.1478y30.41160.39770.5538-0.0414-0.01160.33100.0388y40.25680.45500.33480.25070.12570.22650.0998y5-0.4709-0.0488-0.4802-0.1007-0.0132-0.29390.1923表7.3CanonicalCorrelationsCanCorrAdj.CanCorrApproxStd.ErrorCanRsq10.8487080.7976840.0459820.72030520.7029630.5825560.0831600.49415730.646784.0.0956260.41832940.3542120.1778850.1437730.12546650.268706.0.1525290.072203由表7.2相关系数阵、表7.3典型相关系数得知,第一典型相关系数为0.848708,修正值为0.797684,标准误差为0.045982,典型相关系数的平方为0.720305;第二典型相关系数为0.702963,修正值为0.582556,标准误差为0.083160,典型相关系数的平方为0.494157;第三典型相关系数为0.646784,标准误差为0.095626,典型相关系数的平方为0.418329;第四典型相关系数为0.354212,修正值为0.177885,标准误差为0.143773,典型相关系数的平方为0.125466;第五典型相关系数为0.268706,标准误差为0.152529,典型相关系数的平方为0.072203。表7.4EigenvaluesEigenvalueDifferenceProportionCumulative12.57531.59840.57320.573220.97690.25770.21740.790730.71920.57570.16010.950740.14350.06560.03190.982750.07780.01731.0000由表7.4特征根可以得到特征根、相邻两特征根之差、特征根所占方差信息量的比例、累计方差信息量的比例,其中前三对典型变量所能解释的变异占总变异的95.07%,其他两个典型相关变量的作用很小,一共只解释了总变异的4.93%,因此不予考虑。表7.5:TestofH0:CanCorr[j]=0,j=KKL.RatioApproxFNumDFDenDFPrF10.0667742.884235111.8019.000120.2387392.01822495.40180.008930.4719621.61911577.69710.087840.8113900.7987858.00000.606150.9277970.7782330.00000.5154应用似然比法检验典型相关系数与零的差别,由表7.5检验典型相关系数与零的差别得到p值依次为0.0001、0.0089、0.0878,因此,对于前两组典型相关变量,拒绝小于此对典型变量典型相关系数的所有典型相关系数为0的原假设,因此,前两组变量的相关性的研究可转化为研究前两对典型相关变量的相关性。表7.6Correlations(Structure)VariableCY1CY2CX1CX2y1-0.71220.1036-0.60440.0728y20.72120.56270.61210.3955y30.76370.23260.64820.1635y40.57650.16610.48930.1167y5-0.68030.2445-0.57740.1719x10.5707-0.16840.6725-0.2395x20.41780.45670.49230.6496x30.64170.17420.75610.2479x40.23020.00030.27120.0004x50.1830-0.47690.2156-0.6784x60.29870.09080.35190.1292x70.08930.19440.10520.2765由表7.6典型相关结构,从相关系数判断,体力测试指标除x4(0.2712)、x5(0.2156)、x7(0.1052)外各变量与第一典型相关变量间的相关性比较高,运动能力测试的指标与第一典型相关变量间的相关性都比较高;x5与第二典型相关变量间的相关性比较高,说明第一对典型相关变量对台阶测试(x5)的解释作用不大。从体力测试指标组的变量与运动能力测试的指标组的典型变量之间,运动能力测试的指标组的变量与体力测试指标组的典型变量之间的相关系数可见,各组变量与前两对典型变量之间均有较强的相关性。表7.7StdCanonicalYCoefficientsVariableCY1CY2y1-0.5340160.655116y20.1474291.070199y30.3331930.371059y4-0.0323250.085995y5-0.4079430.938190表7.8StdCanonicalXCoefficientsVariableCX1CX2x10.449775-0.150127x20.1651090.582368x30.6594610.013959x40.067975-0.131520x50.267853-0.723637x60.113355-0.130192x70.0147380.391286由表7.7、表7.8标准化变量的典型变量的系数可知,来自运动能力测试的指标的第一典型变量CY1为:****15432407943.0032325.0333193.0147529.0534016.01yyyyyCYCY1在y4*上的系数近似为0,在y1*、y4*、y5*上的系数为负值,在y3*上的系数较大,因此CY1主要代表了投球等指标。来自体力测试指标的第一典型变量CX1为:*7*6*5*4*3*2*1014738.0113355.0267853.0067975.0659461.0165109.0449775.01xxxxxxxCXCX1在x3*上的系数最大,在x1*上的系数较大,在其余变量上的系数均较小,因此,CX1主要代表了反复横向跳、背力等指标。表7.9StdVariance(