基于模糊聚类的改进的模糊关联规则挖掘算法(英文)

An Improvement of Fuzzy Association Rules Mining Algorithm Based on Fuzzy Clustering

文中提出一种新的方法通过使用模糊C均值对原始数据集进行预处理操作,通过这个操作可以把定量属性值转换为二进制值,继而就会得到原始数据集的模糊版本(由模糊记录和模糊属性组成)。另外,文中又提出了一种基于模糊Apriori算法的快速提取规则的算法,这种算法是利用模糊聚类从先前得到的原始数据集的模糊版本中提取模糊频繁项集从而可以得到模糊关联规则。在文章的最后,实验结果显示了提出的新算法在处理大型数据集时在挖掘时间上要优于传统的Apriori算法。对大型数据库来说,该算法在实用性和可用性上面都有很好的发展前景。

In this paper, propose a methodology by doing pre-processing the original dataset using FCM which can convert quantitative values of attributes to binary values, and then get a fuzzy version I with fuzzy records and fuzzy attributes } of the original dataset. Moreo- ver,prasent a fast algorithm based on the fuzzy Apriori algorithm for rule extraction utilizing fuzzy clustering （ FAFC ） for extracting fuzzy frequent itemsets and fuzzy association rules from the fuzzy version of the original dataset. Eventually, experiments show that the FAFC algorithm outperforms the traditional Apriori algorithm on computing time for huge database. And for huge dataset, the algorithm presented in this paper is found to be promising in terms of practicability and availability.