中文摘要：

传统的系统聚类分析方法不能自动判断结果中适宜的类别数，本文应用两步聚类法与层次聚类法相结合的方式对25种汽车样本进行聚类分析，并利用类型内标准差与总体标准差的比较对聚类效果进行了检验。结果表明，使用这种组合聚类分析方式先科学、客观地确定聚类数目，再进行聚类分析，得到的三类数据样本集合，样本集合间各指标存在显著差异，所得到的结果与客观实际相符，证明了这种新聚类方式的可行性。

英文摘要：

The traditional cluster analysis method can not automatically determine the appropriate clustering num- ber. In this paper the combination method of two-step cluster analysis and hierarchical cluster analysis is applied to an- alyze 25 kinds of automobile samples. And the clustering effect is also examined through the comparison of the interior standard deviation and overall standard deviation. The results show that firstly the combination method determines the clustering number scientifically and objectively, and then three types of data sample collection ＂are obtained by the clus- tering analysis in which the various indicators in the sample collection are significantly different. It can be found that the obtained results basically correspond with the objective reality. And it proves the feasibility of new clustering method.