论文部分内容阅读
将信息熵理论与直觉模糊粗糙集结合起来,提出一种基于互信息的直觉模糊粗糙集属性约简新算法.给出了在直觉模糊环境下,基于互信息的属性重要度和属性依赖度的度量准则.本文所提出的算法以属性重要度和依赖度为双重度量标准,采取可增可删的双向回归算法,在保持分类精度不变的情况下,最后得到决策表的最小属性约简.实例表明在多属性的决策表约简中,在本文提出的算法得到的属性约简的基础上而得到的决策规则是较简捷、较完备的.
Combining the theory of information entropy and intuitionistic fuzzy rough set, a new attribute reduction algorithm of intuitionistic fuzzy rough set based on mutual information is proposed.It is given that in the intuitionistic fuzzy environment, attribute importance and attribute dependence based on mutual information MEASUREMENT PRINCIPLE The algorithm proposed in this paper takes attribute importance and degree of dependence as double measures, adopts bidirectional regression algorithm which can increase or decrease, and finally obtains the minimum attribute reduction of decision table while keeping the classification accuracy unchanged. The examples show that the decision rules obtained on the basis of the attribute reduction obtained by the proposed algorithm are simple and complete in the multi-attribute decision table reduction.