An improved attribute reduction algorithm based on mutual information with rough sets theory

Yang Su-min, Meng Jie, Liu Qi-

Abstract

There aren’t core attributes in some information system, but the core attributes are the basis of attribute reduction algorithm based on mutual information, in order to solve the problem of a new attribute importance degree one new method on the basis of mutual information is proposed in the paper, which consists of its own information entropy and mutual information. Then the corresponding heuristic reduction algorithm is proposed. Experimental results show that the algorithm can solve non-core information system attribute reduction, but also can get attribute reduction faster, and the reduction number is also relatively small.