数据挖掘中的聚类方法

从空间X中给定一个有限的取样点集(或从数据库中取得有限例子的集合),{x~i}_(i 1)~M,聚类的目标是将数据聚集成类,使得类间的相似性尽量小,而类内的相似性尽量大。分类问题(监督)和聚类问题根本的不同是:分类问题中,我们知道训练例的分类属性值,而在聚类问题中,就需要我们在训练例中找到这个分类属性值。1 数据挖掘领域中的聚类研究

Clustering Method in Data Mining

In this paper we introduce clustering method at Data Mining.Clustering has been studied very deeply.In the field of Data Mining,clustering is facing the new situation.We summarize the major clustering methods and introduce four kinds of clustering method that have been used broadly in Data Mitring.Finally we draw a conclusion that the partitional clustering method based on distance in data mining is a typical two phase iteration process1)appoint cluster;2)update the center of cluster.