Rough Set Theory Analysis on Decision Subdivision

Abstract

The degree of subdivision of the decision attribute value influences upon the accuracy of approximation classification, the approximation quality of rules, the core attributes and the information entropy in decision systems based on rough set theory. The finer the decision attribute discretization of a decision table is, the less the accuracy of approximation classification, the approximation quality of rules, and information entropy are on any condition attribute set. Meanwhile, if the attribute values of decision attributes are divided into finer values, then the core attributes set obtained from the finer decision table must include the core attributes set obtained from the previous decision table. These conclusions are proved theoretically. So the discrete degree of decision attributes should be chosen properly. The research is helpful to attribute reduction and enhancing confidences of decision rules.