數(shù)據(jù)分析驗證性實驗報告

數(shù)據(jù)分析驗證性實驗報告題目1、1991年我國30個省、區(qū)、市城鎮(zhèn)居民月平均消費八個指標（單位均為元/人）X1:人均糧食支出X2:人均副食支出 X3:人均煙茶支出X4:人均其它副食支出X5:人均衣著商品支出X6:人均日用品支出X7:人均燃料支出X8:人均非商品支出省區(qū)市X1X2X3X4X5X6X7X8山西8.3523.537.518.6217.4210.001.0411.21內(nèi)蒙古9.2523.756.619.1917.7710.481.7210.51吉林8.1930.504.729.7816.287.602.5210.32黑龍江7.7329.205.429.4319.298.492.5210.00河南9.4227.938.208.1416.179.421.559.76甘肅9.1627.989.019.3215.999.101.8211.35青海10.0628.6410.5210.0516.188.391.9610.81河北9.0928.127.409.6217.2611.122.4912.56陜西9.4128.205.7710.8016.3611.561.5312.17寧夏8.7028.127.2110.5319.4513.301.6611.96新疆6.9329.854.549.4916.6210.651.8813.61湖北8.6736.057.317.7516.6711.682.8312.88云南9.9837.697.018.9416.1511.080.8311.67湖南6.7738.696.018.8214.7911.441.7413.23安徽8.1437.759.618.4913.159.761.2811.28貴州7.6735.718.048.3115.137.761.4113.25遼寧7.9039.778.4912.9419.2711.052.0413.29四川7.1840.917.328.9417.6012.751.1414.08山東8.8233.707.5910.9818.8214.731.7810.10江西6.2535.024.726.2810.037.151.9310.39福建10.6052.417.709.9812.5311.702.3114.69廣西7.2752.653.849.1613.0315.261.9814.57海南13.44555.855.507.459.559.522.2116.30天津10.8544.687.3214．5117.1312.081.2611.57江蘇7.2145.797.6610.3616.5612.862.2511.69浙江7.6850.3711.3513.3019.2514.592.7514.87北京7.7848.448.0020.5122.1215.731.1516.61西藏7.9439.6520.9720.8222.5212.411.757.90上海8.2864.348.0022.2220.0615.120.7222.89廣東12.4776.395.5211.2414.5222.005.4625.50設(shè)前20個省份為第1類G1，21-27號省份（即福建，…，北京）為第2類G2，最后三個省份（西藏、上海、廣東）待判。試判別西藏、上海、廣東各屬哪一類，并計算判別率的回代估計。80二、程序1.錄入數(shù)據(jù)2.按“Analyze→Classify→Discriminant”順序，打開DiscriminantAnalysis主對話框，選擇“類別”為GroupingVariable（分組變量），定義“類別”的區(qū)域，選擇x1、x2、x3、x4、x5、x6、x7、x8為IndepentVariable（解釋變量）。3.點擊Statistics按鈕，進入Statistics對話框，在Descriptives欄選擇Mean（對各組的各個變量作均值和標準差的描述；②在FunctionCoefficients欄（判別函數(shù)的系數(shù)）選擇Fisher’s(Fisher線性判別函數(shù))和Unstandardized（判別方程的非標準化系數(shù)）非標準化函數(shù)；③在Matrices欄選Within-groupscorrelation（組內(nèi)相關(guān)矩陣）,Within-groupscovariance(組內(nèi)協(xié)方差矩陣)，Separate-groupscovariance(組間協(xié)方差矩陣)，Totalcovariance(總協(xié)方差矩陣)。4.點擊Classification按鈕，進入Classification對話框。在PriorProbabilities欄選擇Allgroupsequal；②在display欄選擇casewiseresults（每個個體的結(jié)果），Summerrytable（綜合表）5.點擊Save按鈕，保存選項中可以選擇預測的分類、判別得分以及所屬類別的概率。運行結(jié)果表1協(xié)方差矩陣a類別x1x2x3x4x5x6x7x81x11.142-2.486.886.432.809.328-.049-.512x2-2.48628.166.604-.289-2.9891.861-.3533.573x3.886.6042.639.407.339.204-.208-.027x4.432-.289.4071.9082.1671.204.069.277x5.809-2.989.3392.1674.8622.371.167.260x6.3281.861.2041.2042.3713.920-.176.894x7-.049-.353-.208.069.167-.176.275-.073x8-.5123.573-.027.277.260.894-.0731.8202x15.7884.060-1.377-3.903-6.551-4.759-.047.843x24.06015.941-3.326-9.715-11.772-2.6181.0695.911x3-1.377-3.3265.3974.6406.3731.030.389-.242x4-3.903-9.7154.64019.28617.2255.855-1.6721.189x5-6.551-11.7726.37317.22518.5996.866-.976-.593x6-4.759-2.6181.0305.8556.8664.946-.275.535x7-.0471.069.389-1.672-.976-.275.339.062x8.8435.911-.2421.189-.593.535.0624.024合計x12.3242.272.362-.083-1.058-.380-.011.277x22.27288.673.3487.799-7.70110.541.71913.319x3.362.3483.1811.4751.689.489-.055.022x4-.0837.7991.4757.4765.1133.789-.2191.964x5-1.058-7.7011.6895.1137.9672.892-.135-.353x6-.38010.541.4893.7892.8925.493-.0822.196x7-.011.719-.055-.219-.135-.082.287.067x8.27713.319.0221.964-.3532.196.0673.614a.總的協(xié)方差矩陣的自由度為26。該表給出各類的協(xié)方差矩陣和總協(xié)方差矩陣。表2：有關(guān)典型判別函數(shù)的輸出表特征值函數(shù)特征值方差的%累積%正則相關(guān)性14.937a100.0100.0.912a.分析中使用了前1個典型判別式函數(shù)。Wilks的Lambda函數(shù)檢驗Wilks的Lambda卡方dfSig.1.16837.4048.000標準化的典型判別式函數(shù)系數(shù)函數(shù)1x1.376x2.891x3-.118x41.006x5-.644x6.393x7.372x8-.310結(jié)構(gòu)矩陣函數(shù)1x2.733x8.349x6.274x4.238x1.120x7.078x5-.056x3.021判別變量和標準化典型判別式函數(shù)之間的匯聚組間相關(guān)性按函數(shù)內(nèi)相關(guān)性的絕對大小排序的變量。組質(zhì)心處的函數(shù)類別函數(shù)11-1.26523.614在組均值處評估的非標準化典型判別式函數(shù)分類處理摘要已處理的27已排除的缺失或越界組代碼0至少一個缺失判別變量0用于輸出中27組的先驗概率類別先驗用于分析的案例未加權(quán)的已加權(quán)的1.5002020.0002.50077.000合計1.0002727.000解：由上可得：分類函數(shù)系數(shù)類別12x16.8308.051x22.1322.997x3-.395-.710x4-2.117-.126x55.9234.822x6-.236.703x77.08910.460x82.2801.294(常量)-119.594-171.802Fisher的線性判別式函數(shù)(1)、判別函數(shù)表達式為：F2=8.051x1+2.997x2-0.710x3-0.126x4+4.822x5+0.703x6+10.466x7+1.294x8-171.802F1=6.836x1+2.132x2-0.395x3-2.117x4+5.923x5-0.236x6+7.089x7+2.280x8-119.594協(xié)方差S1矩陣為：1x11.142-2.486.886.432.809.328-.049-.512x2-2.48628.166.604-.289-2.9891.861-.3533.573x3.886.6042.639.407.339.204-.208-.027x4.432-.289.4071.9082.1671.204.069.277x5.809-2.989.3392.1674.8622.371.167.260x6.3281.861.2041.2042.3713.920-.176.894x7-.049-.353-.208.069.167-.176.275-.073x8-.5123.573-.027.277.260.894-.0731.820協(xié)方差S2矩陣為協(xié)方差矩陣s2a類別x1x2x3x4x5x6x7x82x15.7884.060-1.377-3.903-6.551-4.759-.047.843x24.06015.941-3.326-9.715-11.772-2.6181.0695.911x3-1.377-3.3265.3974.6406.3731.030.389-.242x4-3.903-9.7154.64019.28617.2255.855-1.6721.189x5-6.551-11.7726.37317.22518.5996.866-.976-.593x6-4.759-2.6181.0305.8556.8664.946-.275.535x7-.0471.069.389-1.672-.976-.275.339.062x8.8435.911-.2421.189-.593.535.0624.024(2)判別準則：若F1>F2，則；若F1<F2，則；(3)對已知類別樣品判別分類樣品號原類號判歸類別111211311411511611711811911101111111211131114111511161117111811191120