Attribute Reduction Algorithm of Continuous Domain Decision Table Based on Similar Matrix

Abstract

Combining fuzzy set with rough set, attribute reduction algorithm of continuous domain decision table is studied. First, continuous attribute value are transformed into fuzzy value with triangular membership function. and then, similarity degree of two fuzzy objects and similarity class of each fuzzy object are defined, in the meantime, characteristic vector of continuous attribute which is made up of similarity class of each fuzzy object is provided. Next, digital characteristic vector of continuous attribute is presented and similar matrix of continuous attributes is proposed. Finally, a new attribute reduction algorithm is provided. Also, the new algorithm is verified through an illustrative example.