基于粒关系包含度矩阵的属性约简算法研究  

Research on Attribute Reduction Algorithm Based on Inclusion Degree Matrix of the Granular-Relation

作　　者：陈志恩[1] 马旭[1] 

机构地区：[1]宁夏师范学院数学与计算机科学系

出　　处：《数学的实践与认识》2018年第5期134-140,共7页Mathematics in Practice and Theory

基　　金：宁夏高校科学研究基金项目(NGY2016172);宁夏科技厅软科学研究项目;宁夏师范学院科研基金项目(NXSFZD1712);“十三五”宁夏高校教学名师基金项目(nxgxjxms20174202019);宁夏高等学校一流学科建设资助项目(NXYLXK2017B11)

摘　　要：针对信息系统属性约简问题，通过借助粒关系包含度矩阵这一中间工具，给出一种决策表属性启发式约简算法．首先，计算决策表中条件属性与决策属性之间的粒关系包含度矩阵；然后，将粒关系包含度矩阵中隐含的信息LB作为启发式算子对决策表进行属性约简；最后，删除冗余属性并设置终止条件，实现决策表的属性约简．通过实例验证了该算法的有效性．Aiming at the problem of attribute reduction of decision information system, A heuristic attribute reduction algorithm based on the decision table was proposed by means of the inclusion degree matrices of the granular-relation which is a intermediate tool. Firstly, the inclusion degree of the granular-relation matrices between the condition attributes and the decision attributes was calculated. Then the attribute was reduced according to Lswhich is hidden in the inclusion degree of the granular-relation matrices, Finally, redundant attributes were removed and the termination condition was seted, and the attribute reduction of decision tablerealized. Experimental result showed that the method proposed in this paper is effective.

关 键 词：决策表 包含度 粒关系矩阵 属性约简 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]