计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (28): 135-138.

• 数据库、信号与信息处理 • 上一篇    下一篇

基于聚类率的决策表属性约简

路  静,张  涛,任宏雷   

  1. 燕山大学 信息科学与工程学院,河北 秦皇岛 066004
  • 出版日期:2012-10-01 发布日期:2012-09-29

Reduction of attribute in decision table based on clustering rate

LU Jing, ZHANG Tao, REN Honglei   

  1. College of Information Engineering, Yanshan University, Qinhuangdao, Hebei 066004, China
  • Online:2012-10-01 Published:2012-09-29

摘要: 根据经典粗糙集方法,在对可约简决策表进行属性约简时可能出现核不存在无法找到起点属性从而无法约简的问题。针对该问题,提出了基于聚类率的属性约简方法。计算决策表的区分矩阵,以属性区分度为基础,在属性区分度相同的情况下,利用聚类率修正属性重要度,保证起点属性存在的必然性,从而完成起点属性的求取并获得决策表的属性约简。实验分析表明,方法可以保证可约简决策表中起点属性的计算,且利用该方法获得的约简结果保持了较高的决策准确率,是有效可行的。

关键词: 粗糙集, 决策表, 聚类率, 属性约简

Abstract: According to classical rough set, when we reduce the attribute of a decision table , it may appear core does not exist that couldn’t find a starting point attribute and be unable to reduction problem. Aiming at this issue, this paper proposes a kind of methodbased on clustering rate of the attribute reduction. This method firstly calculates decision table of the discernibility matrix, attribute to differentiate as the foundation, in the attribute of the same differentiate conditions, the clustering rate fixed attribute importance, guarantee the starting point attribute necessity of existence, so it can find the starting point and win the attributes of the decision table attribute reduction. Experimental results show that the proposed method can make gain the starting pointattribute, and by using the methods of obtaining the reduction results maintains high decision accuracy, and it is effective.

Key words: rough set theory, decision table, clustering rate, attribute reduction