在SPSS中基于相关系数矩阵变量的聚类分析  

Cluster Analysis of the Correlation Coefficient Matrix Based Variables in SPSS

作　　者：朱红兵[1] 席凯强[1] 刘建通[1] 

机构地区：[1]首都体育学院,北京100191

出　　处：《首都体育学院学报》2014年第6期568-571,共4页Journal of Capital University of Physical Education and Sports

摘　　要：在体育文献中,由于刊登原始数据需占用大量的篇幅,因此,很难在体育文献中见到多变量的原始数据资料,所以,能在文献资料中查阅到的描述变量之间线性关联程度的主要资料即相关系数矩阵。在SPSS中,变量聚类分析是建立在原始数据的基础上的,没有专门的分析模块来对只有相关系数矩阵的资料进行统计分析,因此,有必要对此展开研究。研究表明,在只有相关系数矩阵,而没有原始数据信息时,可按SPSS中提供的建立相关系数矩阵的固定格式在SPSS中建立相关系数矩阵的数据文件,在聚类分析之前需对相关系数矩阵进行转换,使之变成等价的距离矩阵,则在调用CLUSTER过程时,可以得到正确的树形聚类输出图,从而实现变量的聚类分析。Since the publication of raw data takes up a lot of space, raw data of multivariable are rarely found in sport literature. Therefore, most data that describes the linear correlation between variables to be searched from literature is the correlation coefficient matrix. Because there is no particular module that specialized in statistical analysis of correlation coefficient matrix data in SPSS and also cluster analysis of variables is built up from raw data, it is necessary to research on this topic. Studies have shown that, with only the correlation coefficient matrix data and without raw data, we can set up correlation coefficient ma- trix data files according to the fixed format of setting up correlation coefficient matrix provided in SPSS. Prior to cluster analysis, it is essential to transform the correlation coefficient matrix to an equivalent dis tance matrix using so that we can obtain a proper cluster dendrogram plot in order to achieve cluster anal- vsis.

关 键 词：SPSS 相关系数 矩阵 聚类分析 

分 类 号：G80-32[文化科学—运动人体科学]