fisher判別.doc

實(shí)驗(yàn)名稱：fisher判別一、實(shí)驗(yàn)?zāi)康暮鸵笸ㄟ^(guò)上機(jī)操作，完成spss軟件的fisher判別二、實(shí)驗(yàn)內(nèi)容和步驟依次點(diǎn)擊，選擇discriminant如下圖所示進(jìn)行操作點(diǎn)擊ststistics,進(jìn)行以下操作點(diǎn)擊classification，進(jìn)行以下操作點(diǎn)擊save，進(jìn)行以下操作Analysis Case Processing SummaryUnweighted CasesNPercentValid15100.0ExcludedMissing or out-of-range group codes0.0At least one missing discriminating variable0.0Both missing or out-of-range group codes and at least one missing discriminating variable0.0Total0.0Total15100.0該表為分析案例處理摘要表，反映的是有效樣本與缺失值的情況，可以看出本案例中有效值有15個(gè)，缺失值為15個(gè)Group Statistics類別MeanStd. DeviationValid N (listwise)UnweightedWeighted1.00X1188.600057.1384355.000X2150.400016.5015255.000X313.80005.9329655.000X420.000013.3229155.0002.00X1157.000041.1703855.000X2115.000014.8155355.000X37.00001.8708355.000X413.60007.5365855.0003.00X1151.000033.8008955.000X2121.400013.0115355.000X35.00001.8708355.000X48.00007.3143755.000TotalX1165.533345.110761515.000X2128.933321.049151515.000X38.60005.220841515.000X413.866710.391391515.000該表為分組統(tǒng)計(jì)量表，是各組變量的描述統(tǒng)計(jì)分析，從表中可以看出各個(gè)類別的均值，標(biāo)準(zhǔn)差。X1在3組中差別較大Tests of Equality of Group MeansWilks LambdaFdf1df2Sig.X1.8571.003212.396X2.4268.074212.006X3.4427.564212.007X4.7621.879212.195該組為組均值的均等性檢驗(yàn)，從表中可以看出 X2.X3在兩組中的均值差異是顯著的。Test ResultsBoxs M45.046FApprox.1.089df120df2516.896Sig.357Tests null hypothesis of equal population covariance matrices.上表為協(xié)方差矩陣的均等性檢驗(yàn)，是對(duì)3組進(jìn)行的總體協(xié)方差矩陣是否相等的統(tǒng)計(jì)檢驗(yàn)，因?yàn)閜值=0.357，所以可知3組的協(xié)方差矩陣差異不顯著。EigenvaluesFunctionEigenvalue% of VarianceCumulative %Canonical Correlation13.044a93.693.6.8682.207a6.4100.0.414a. First 2 canonical discriminant functions were used in the analysis.Wilks LambdaTest of Function(s)Wilks LambdaChi-squaredfSig.1 through 2.20516.6498.0342.8281.9783.577上表是典型判別函數(shù)的分析結(jié)果，其中第一個(gè)表中反映了判別函數(shù)的特征值分別為3.044a 0.207a 分別解釋了93.6%與6.4%的方差。Standardized Canonical Discriminant Function CoefficientsFunction12X1.453-.175X2.596-.811X3.662.600X4.299.608上表是標(biāo)準(zhǔn)化典型判別函數(shù)，反映的是標(biāo)準(zhǔn)化的判別函數(shù)。Canonical Discriminant Function CoefficientsFunction12X1.010-.004X2.040-.055X3.176.160X4.031.062(Constant)-8.7845.448Unstandardized coefficients上表為未標(biāo)準(zhǔn)化的典型判別函數(shù)系數(shù)，即費(fèi)希爾判別函數(shù)系數(shù)，可以將樣本觀測(cè)值直接帶入函數(shù)以求得判別函數(shù)Casewise StatisticsCase NumberActual GroupHighest GroupSecond Highest GroupDiscriminant ScoresPredicted GroupP(Dd | G=g)P(G=g | D=d)Squared Mahalanobis Distance to CentroidGroupP(G=g | D=d)Squared Mahalanobis Distance to CentroidFunction 1Function 2pdfOriginal111.4292.9981.6932.00213.9872.7551.127211.4192.9771.7392.0219.4462.0471.261311.5752.9991.1072.00016.4282.877-.854413*.1002.5784.5971.2656.153.430-1.788511.7882.999.4752.00114.8712.886.008622.4022.4571.8203.4032.072.161-.264722.4382.7001.6503.3003.347-2.0351.186823*.9032.518.2052.480.356-1.158-.032922.4162.6621.7541.1694.479.323.9321022.5652.6161.1423.3842.084-1.971.7861132*.9912.616.0173.379.987-.842.4281233.8692.681.2802.3141.833-.873-.8311333.2362.8392.8842.1446.407-.447-1.9621433.4632.5871.5402.4132.240-2.369.0