An Attribute Reduction Algorithm Based on Clustering and Attribute-Activity Sorting

Attribute reduction is one of key processes in rough set theory. In this paper, a new attribute reduction algorithm and a definition of Attribute-Activity is proposed with theoretical basis. It uses Attribute-Activity to quantify the partition capability for an attribute and makes a rough sorting, then makes clustering analysis by calculating the similarity among attributes to modify the sorting to obtain a sequence indicating the attribute significance, finally obtain a better attribute reduction. In addition, a contrastive analysis of efficiency and feasibility in new algorithm and other traditional algorithms is in detailed, shows that this algorithm is effective.