华东理工大学多元统计分析与SPSS应用实验(5)

1华东理工大学2013—2014学年第二学期《多元统计分析与SPSS应用》实验报告5班级学号姓名开课学院商学院任课教师任飞成绩实验内容：实验5聚类分析方法1.熟悉SPSS中聚类分析的距离选择功能AnalyzeClassifyHierarchicalCluster在Measure栏中选择距离测度方法：Block---dij(1)Euclideandistance---dij(2)SquaredEuclideandistanceChebychev---dij(∞)Minkowski---dij(q)Cosine---cij(1)Pearsoncorrelation---cij(2)2.熟悉SPSS中聚类分析的系统聚类功能AnalyzeClassifyHierarchicalCluster在ClusterMethod栏中选择系统聚类方法：最短距离法(NearestLinkage)最长距离法(FurthestLinkage)中间距离法(MedianLinkage)重心法(Centroidclustering)类平均法(Between-groupsLinkage)(Within-groupsLinkage)离差平方和法(Ward’smethod)实验要求：1.选用例题5.1文件中的变量，完成系统聚类法的各种结果的比较分析例题5.1从21个工厂各抽一件同类产品，每个产品测两个质量指标，记作x1、x2，要求将各厂的产品按质量情况进行分类。2.对案例：Crop’Pain连锁店，展开聚类分析讨论思考：分类完成后，能进行哪些统计分析？改革二十多年来，中国经济的发展阶段划分的思考2教师评语：教师签名：年月日实验报告：5.11、打开例题5.1.sav文件，，如图5.1.1所示依次Analyze→Classify→HierarchicalCluster,将x1，x2选入Variable框中图5.1.12、单击Statistics，选择ProximityMatrix，ClusterMembership中选择RangeofSolutions，依次输入6,12，如图5.1.2，单击Method，ClusterMethod为默认组间平均链锁法，单击Interval选项，激活右侧的参数框，单击下拉箭头，共有8个选项，选择Euclideandistance（欧式距离），如图5.1.3，单击Continue，单击Plots，选择Dendrogram，如图5.1.4单击Continue，选择OK得到结果图5.1.23图5.1.3图5.1.44图5.1.5距离矩阵ProximityMatrix.0001.0002.2363.6064.4725.0007.0717.2117.8109.2205.0004.4725.0006.7086.4035.0997.0008.0007.0719.05511.4021.000.0002.0002.8284.1234.4726.4036.7087.2118.6024.4723.6064.2435.8315.8314.1236.0007.0006.0838.06210.4402.2362.000.0002.0002.2362.8285.0005.0005.6577.0716.3255.0005.8317.0717.6164.1236.3257.2806.7088.54411.1803.6062.8282.000.0002.2362.0003.6064.1234.4725.8316.0004.1235.0995.8317.0712.2364.4725.3855.0006.7089.4344.4724.1232.2362.236.0001.0003.1622.8283.6065.0008.0626.3257.2808.0629.2204.2436.4037.2117.0718.60211.4025.0004.4722.8282.0001.000.0002.2362.2362.8284.2438.0006.0837.0717.6169.0553.6065.6576.4036.4037.81010.6307.0716.4035.0003.6063.1622.236.0001.4141.0002.2369.2207.0718.0628.06210.0504.0005.3855.8316.3257.21110.0007.2116.7085.0004.1232.8282.2361.414.0001.0002.23610.0508.0009.0009.22011.0005.0996.7087.2117.6168.60211.4027.8107.2115.6574.4723.6062.8281.0001.000.0001.41410.1988.0629.0559.05511.0455.0006.3256.7087.2808.06210.8179.2208.6027.0715.8315.0004.2432.2362.2361.414.00011.4029.22010.19810.00012.1666.0837.0717.2808.0628.54411.1805.0004.4726.3256.0008.0628.0009.22010.05010.19811.402.0002.2361.4143.1621.4145.3855.6576.4035.0006.7088.0624.4723.6065.0004.1236.3256.0837.0718.0008.0629.2202.236.0001.0002.2363.0003.1623.6064.4723.1625.0997.0715.0004.2435.8315.0997.2807.0718.0629.0009.05510.1981.4141.000.0002.0002.0004.1234.2435.0003.6065.3857.0006.7085.8317.0715.8318.0627.6168.0629.2209.05510.0003.1622.2362.000.0002.8284.1233.1623.6062.2363.6065.0006.4035.8317.6167.0719.2209.05510.05011.00011.04512.1661.4143.0002.0002.828.0006.0835.8316.4035.0006.4037.2805.0994.1234.1232.2364.2433.6064.0005.0995.0006.0835.3853.1624.1234.1236.083.0002.2363.1622.8284.4727.2117.0006.0006.3254.4726.4035.6575.3856.7086.3257.0715.6573.6064.2433.1625.8312.236.0001.0001.0002.2365.0008.0007.0007.2805.3857.2116.4035.8317.2116.7087.2806.4034.4725.0003.6066.4033.1621.000.0001.4141.4144.2437.0716.0836.7085.0007.0716.4036.3257.6167.2808.0625.0003.1623.6062.2365.0002.8281.0001.414.0002.0004.4729.0558.0628.5446.7088.6027.8107.2118.6028.0628.5446.7085.0995.3853.6066.4034.4722.2361.4142.000.0002.82811.40210.44011.1809.43411.40210.63010.00011.40210.81711.1808.0627.0717.0005.0007.2807.2115.0004.2434.4722.828.000Case123456789101112131415161718192021123456789101112131415161718192021EuclideanDistanceThisisadissimilaritymatrix图5.1.6凝聚状态表AgglomerationSchedule17191.00000612131.0000012891.000007561.0000013121.000001617181.207109781.207031011151.414001417201.88360157101.9627018342.000001312142.1182014352.3251141611122.4408121716173.1750917133.5925131811164.699141519175.29716102011215.817170201117.34918190Stage1234567891011121314151617181920Cluster1Cluster2ClusterCombinedCoefficientsCluster1Cluster2StageClusterFirstAppearsNextStage5图5.1.7类成员聚类表ClusterMembership111111111111112222222332222244332224433222554433355443335544333654433376554448766544876654498765447655444109876551110987651110987651110987651110987651211109876Case12345678910111213141516171819202112Clusters11Clusters10Clusters9Clusters8Clusters7Clusters6Clusters图5.1.8树形图6结果分析：本次聚类分析采取变量之间距离的计算用欧式距离，类与类之间的距离采用组间平均链锁法，将21个样本分成6~12组比较，图5.1.5为距离矩阵，给出了各样本之间的欧式距离；图5.1.6为凝聚状态表，第一列表示聚类分析的第几步，第二、三列表示本步骤中哪两个个案或者小类聚成一类。第四列是个案距离或者小类距离。第五列、第六列表示参与本步骤聚类的是个案还是小类，0表示个案，非零表示小类，具体数字表示第几步生成的小类。第七列标志本步骤的结果将在第几步中用到。图5.1.7为类成员聚类表，给出了分别聚成6~12的最终聚类结果，例如当指定聚类成6类时，1,2聚为一类，3~6聚为一类，7~10聚为一类，11~15聚为一类，16~20聚为一类。图5.1.8为聚类树形图，既给出了聚类过程，也给出对应相应类与类之间的距离。5.2依据题意，需对Michel向上司管理报告作出选址建议。首先认为选址地点需要最低满足毛利/投资额=0.26的基本要求。所以聚类分析前先筛选出毛利/投资额=0.26的样本，点击“Data-SelectCases”，选择“Ifconditionissatisfied”，输入“毛利/投资额=0.26”，如图5.2.1，点击“Continue”得到结果如图5.2.2图5.2.17图5.2.23、依次Analyze→Classify→HierarchicalCluster,将投资、店堂面积选入Variable框中，单击Statistics，选择ProximityMatrix，ClusterMembership中选择RangeofSolutions，依次输入3，6；如图5.1.2，单击Method，ClusterMethod选为最短距离法，单击Interval选项，激活右侧的参数框，单击下拉箭头，共有8个选项，选择Euclideandistance（欧式距离），如图5.1.3，单击Continue，单击Plots，选择Dendrogram，如图5.2.3单击Continue，选择OK得到结果图5.2.