Coefficient of variation based decision tree for fuzzy classification

Abstract

This paper considers a decision system with a fuzzy decision attribute (with finite set of values) to account for uncertainty. A novel fuzzy classification approach using a class of decision trees is developed. A decision tree is constructed for each decision category using Coefficient of Variation Gain as the attribute selection measure. A metric based on Residual Sum of Squares (RSS) to compare the fuzzy classifier is presented. The methodology of constructing the classifier and its performance aspects are presented.