●离差平法和法( wards method) d(G, Ga=sp+g-S,-s 反映G,聚集程度 p =>(Xi-X,' (Xi-X, 反映G聚集程度 Sn=∑(Xn-X2)(X-X P+q ∑(X Ip+q Xna(X Ip-+q p+q
离差平法和法（ward’s method） d Gp Gq = Sp+q − Sp − Sq ( , ) 2 = = − − np i Sp Xi p X p Xi p X p 1 ( ) ( ) = = − − nq i Sq Xi q Xq Xi q Xq 1 ( ) ( ) + = + + + + − + = − np nq i Sp q Xi p q X p q Xi p q X p q 1 ( ) ( ) 反映 G p 聚集程度 反映 Gq 聚集程度
聚类方法的SPSS实现 o Analyze→→ Classify K-means cluster(动态聚类法) Hierarchical cluster(系统聚类法)
聚类方法的SPSS实现 Analyze Classify K-means cluster(动态聚类法) Hierarchical cluster(系统聚类法）
对样本分类的SPSS实现 2005社会发展( factor)say- SPSS Data Editor File Edit View Data Transform Analyze Graphs Utilities Window Help 哆回舀國面 Reports Descriptive Statistics 1: region bel] Tables rgon人均GDP|美 Compare Means 高校数「卫生机构 General Linear Model 数 1beijing 3382573 Mixed Models 6770481800 aning 2678270 prelate 4200247200 hebei 1113309Regression 48600104600 4 shanxi 9411298Lgel 5900943000 5]mongolia 1233222Classify TwoStep Cluster 6 liaonian 1433439 Data Reducti K-Means Cluster 100996 Hi erarchical Cluster 8h108939900aret ree g hanghai 3889586 Time Se Discriminant 10 jiangshu 1850077 Survival 114U 324UU Multiple Response 12ah66517221aay68010500 1122072655 8100919700 13 fujian Complex Samples 4035 93400
对样本分类的SPSS实现
Hierarchical cluster窗口 Hierarchical Cluster Analysis ∨AR001 variable[sl VARO0002 命VAR00 Paste 命VAR004 Reset 争VAR05 命vAR Cancel 争VAR07 Label cases b Hel ◆VAR0B Cluster o Cases C variables D v Statistics v Plots Statistics Plots Method say
Hierarchical cluster窗口
Hierarchical Cluster Analysis company Variable[s] OK ∨AR00001 ∨AR002 Paste ④VAR003 Reset 排VAR04 企W4nnnn Cancel Label cases bi Help Cluster fPCases C variables 样本聚类 Displa v Statistics Plots Statistics Method s ave
样本聚类