基于Vague图的最优树聚类

Optimal tree clustering based on Vague diagram

在智能系统的开发和研究中,聚类分析是一个很重要的问题。为了减少传统的基于等价关系的聚类分析方法所造成的数据"失真"程度,提出了Vague图的最优树聚类方法,该方法引入了Vague树的概念,并通过计算最小二分割设计出Vague最优树,将该方法与基于等价关系的聚类分析方法和直接聚类法进行比较分析,结果表明该方法不仅仅有效地减少了"噪声",还具有连续性和全面性的特点,其聚类水平也更加合理。

Clustering analysis is a very important problem in the research and development of intelligence system. In order to mitigate the data ＂distortion＂ degree caused by the traditional clustering analysis based on equivalence relations, the optimal tree Vague graph clustering method is proposed which introduces the concept of Vague trees and designs the Vague optimum tree by caculating the minimum of two segmentation. Compared with clustering analysis based on equivalence relations and directive clustering method, the result shows that the method not only effectively reduces the ＂noise＂, but also has the characteristics of continuity and comprehensiveness and more reasonable level of clustering.