摘要
针对多数属性约简算法在构造差别矩阵时会产生大量重复元素,并且在约简过程中需要频繁的对整个论域进行等价划分等弱点,对降低属性约简和规则生成的复杂度作了研究。利用基于等价类的差别矩阵进行属性重要性计算,从规则相容度的角度出发,提出了一种等价划分域动态缩减的属性约简算法。该算法提高了属性约简的效率,同时完成确定规则决策树的构造。最后给出了具体的实例。
Aiming at the problems that most attribute reduction algorithm can bring in much duplicate element and must work on the whole decision table, the methodology of how to reduce the attribute reduction and rule generation complexity is presented. Firstly, discernibility matrix on equivalence class to calculate the importance of attribute is constructed. Secondly an attribute Reduction algorithm based on dynamic reducing domain is presented, which not only improve the efficiency but also complete the structure of decision tree. Finally, an example is carried out to demonstrate the algorithm.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第9期2030-2033,2158,共5页
Computer Engineering and Design
关键词
粗糙集
差别矩阵
规则相容度
属性约简
动态划分
rough set
discernibility matrix
rule compatibility
attribute reduction
dynamic partitioning
