Abstract: Analyses the defect of distance function which has been introduced in clustering arithmetic before both from similar degree and sustain degree，and puts forward the arithmetic which improves distance function-Mix arithmetic.The research has certified that the arithmetic can decrease the expense in time and the error degree in result of clustering in the process of achievement，and has improved the quality of clustering，can reach the better clustering result accordingly.

Key words: data mining, frequent pattern, clustering, distance function

摘要： 从模式的相似度信息和支持度大小两方面分析了前人聚类算法中采用的距离函数的缺陷，提出了改进距离函数的新算法—Mix算法。实验研究证明，算法在实现过程中可以相应减少时间消耗和聚类结果的错误程度，提高聚类质量，从而得到比较好的聚类效果。

关键词: 数据挖掘, 频繁模式, 聚类, 距离函数