Attribute Reduction Based on the Minimum Hybrid Entropy

Fuzzy set theory and rough set theory are effective tools for dealing with incomplete and inaccurate knowledge of information systems. In this paper, for the difference of fuzzy equivalence relation matrix of the attributes in information system, we establish a minimum hybrid entropy method measuring the relation of the attributes, and propose a new attribute reduction method based on the minimum hybrid entropy. Finally, we make a comparison analysis by a concrete example, the results indicate our method is suitable for large scale database mining, and it possess many interesting advantages of easy operation and small computation complexity, so it can be widely used in many fields and has strong application value.