基于二进制可辨矩阵属性重要度的属性约简算法

粗糙集理论是一个新的数据挖掘方法,是在保持分类能力不变的情况下,利用等价类,通过属性约简和规则约简,达到挖掘知识并简化知识的目的.但属性约简是一个NP难题,需要通过启发式知识实现.文中提出了一种利用二进制可辨矩阵的属性重要度实现属性约简的算法,该算法能快速求最少属性且实现简单,并通过理论和实例证明了其正确性.

Algorithm for Attribute Reduction Based on Attribute Significance of Binary Discernible Matrix

Rough set theory is a new method of data mining.Its basic thought is utilizing equivalence relation class,through attribution reduction and rule reduction,to excavate knowledge and reduce knowledge.But the attribution reduction is a NP problem,and it needs to be realized by knowledge of elicitation method. The algorithm for attribute reduction based on attribute significance of binary discernible matrix is proposed. The algorithm can get the smallest attributes quickly and be realized easily. And it is proved to be workable in the theory and practice.