聚类联合关联规则的数据挖掘技术

The Combining Technology of Data Mining Based on Clustering and Association Rules

摘要: 尽管聚类分析和关联规则作为两个主要应用方法都可以实现数据挖掘功能，但两者存在三大差异，聚类的数据类型为连续型，关联规则为离散型；聚类体现挖掘的描述功能，关联规则体现预测/验证功能；聚类的输出形式为类簇，关联规则输出的是规则。两者同时具有一定的互补性。因此，本文将两者结合起来，先对样本集进行聚类分析，使样本实体获得各自的类别信息；再对这些带有分类属性的样本进行关联规则挖掘，使得挖掘运算有效降维且具有更好的挖掘目标，挖掘结果可以清晰地显示聚类形成的原因和聚类之间的关系等潜在知识。实验表明，本文介绍的联合挖掘技术可以取得更好的挖掘效果，具有很大的实用价值。

Abstract: Although clustering analysis and association rules as two main application methods can achieve data mining, but both two methods have three different. The data type of clustering operation is continuous and association rules are discrete. Clustering reflects the description function of the mining and association rules reflect prediction/validation function. The output form of clustering is clusters, and association rules then output the lines of rule. At the same time, both of them have some complementary to each other. So, this paper combined the both methods. The clustering analysis for the set of samples was first executed. This processing will make samples for their respective category entity information. Then, run association rules mining according to the samples what with classification properties. The method show the potential knowledge further including causes of the formation of clustering and the relationship between clusters. The experiment shows that the mining technology has better effect and great value of application.