摘要： 介绍频繁项集的概念及其性质，把最大频繁项集作为聚类的依据，提出一种基于最大频繁项集的聚类算法，将关联分析与聚类分析相结合，在聚类中充分利用数据项间的关联性，无须输入聚类个数，并在多个数据集上进行实验。实验结果表明，与传统的基于距离的聚类算法K-Means相比，该算法减少计算数据对象间距离的时间花销，提高算法的效率，具有较高的聚类精度，聚类结果的可解释性也
较强。

关键词: 聚类分析, 最大频繁项集, Apriori性质

Abstract: The concept and properties of frequent itemsets are introduced. The maximal frequent itemsets is made as clustering basis. A new Clustering Algorithm Based on Maximal Frequent Itemsets(CABMFI) is proposed, which integrates the association analysis and clustering analysis. The relevance between data items is fully used, and need not input the clustering number. It is tested with several datasets. Experimental results show that, compared with traditional distance-based K-Means clustering algorithm, this algorithm can reduce the time cost of computing the distance of objects, improve the efficiency, and has better accuracy. The interpretability of clustering results is also well.

Key words: clustering analysis, maximal frequent itemsets, Apriori properties

中图分类号: 

• TP301.6