实验报告9 典型相关分析

《实验报告9 典型相关分析》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、实验九典型相关分析实验目的和要求能利用原始数据与相关矩阵、协主差矩阵作相关分析,能根据SAS输出结果选出满足要求的几个典型变量．实验要求:编写程序，结果分析．实验内容：4.9方法一：SASdataexamp4_9;inputx1-x2y1-y2;cards;19115517914519514920115218114818514918315318814917614417114220815719215218915019014919715918915218815219715919215018715117915818614818314717414717415018515219015919

2、5157188151187158163137161130195155183158186153173148181145182146175140165137192154185152174143178147176139176143197167200158190163187150;run;proccancorrdata=examp4_9corr;varx1-x2;withy1-y2;run;TheSASSystem16:48Sunday,October31,20121TheCANCORRProcedureCorrelationsAmongtheOriginalVariablesCorre

3、lationsAmongtheVARVariables（变量x1-x2的相关系数矩阵）x1x2x11.00000.7504x20.75041.0000CorrelationsAmongtheWITHVariables（变量y1-y2的相关系数矩阵）y1y2y11.00000.8397y20.83971.0000CorrelationsBetweentheVARVariablesandtheWITHVariables变量x1-x3与y1-y3的相关系数矩阵y1y2x10.70920.7050x20.71400.7440TheSASSystem16:48Sunday,October3

4、1,20122TheCANCORRProcedureCanonicalCorrelationAnalysisAdjustedApproximateSquaredCanonicalCanonicalStandardCanonicalCorrelationCorrelationErrorCorrelation10.8014200.7881670.0745910.64227420.084522.0.2070250.007144TestofH0:ThecanonicalcorrelationsintheEigenvaluesofInv(E)*Hcurrentrowandallthatfo

5、llowarezero=CanRsq/(1-CanRsq)LikelihoodApproximateEigenvalueDifferenceProportionCumulativeRatioFValueNumDFDenDFPr>F11.79541.78820.99600.99600.355170136.784400.000320.00720.00401.00000.992855970.151210.7014检验假设检验统计量，为第一、第二自由度．由检验结果可知，，．故只有第一对典型变量显著相关．取第一对进行分析即可．MultivariateStatisticsandFApprox

6、imationsS=2M=-0.5N=9StatisticValueFValueNumDFDenDFPr>FWilks'Lambda0.355170136.784400.0003Pillai'sTrace0.649418305.054420.0021Hotelling-LawleyTrace1.802633138.884230.0002Roy'sGreatestRoot1.7954376918.85221<.0001NOTE:FStatisticforRoy'sGreatestRootisanupperbound.NOTE:FStatisticforWilks'Lambdaise

7、xact.TheSASSystem16:48Sunday,October31,20123TheCANCORRProcedureCanonicalCorrelationAnalysisRawCanonicalCoefficientsfortheVARVariablesV1V2x10.0474933802-0.144760311x20.08404777910.1959848929RawCanonicalCoefficientsfortheWITHVariablesW1W2y10.04