Matlab提供了两种方法进行聚类分析。

一种是利用 clusterdata函数对样本数据进行一次聚类，其缺点为可供用户选择的面较窄，不能更改距离的计算方法；

另一种是分步聚类：（1）找到数据集合中变量两两之间的相似性和非相似性，用pdist函数计算变量之间的距离；（2）用 linkage函数定义变量之间的连接；（3）用 cophenetic函数评价聚类信息；（4）用cluster函数创建聚类。

-Matlab provides two methods of cluster analysis. One is a function of the sample data using clusterdata to conduct a cluster, its disadvantages for the user to select a narrower face and can not change the method of calculating the distance　the other is the sub-Buju categories: (1) find a data set variable pairwise similarity between the non-similarity, calculated using pdist function, the distance between variables　(2) function definition with the linkage connection between variables　(3) with the cophenetic clustering evaluation function, the information　(4) with cluster function is to create clustering.