A New Discretization Approach of Continuous Attributes

Abstract

To deal with the discretization problem in an information system, a new discretization approach of continuous attributes is proposed in this paper based on the relative entropy and rough set theory. The candidate interval class-information entropy is used to select the threshold boundary for discretization in this method. And the redundant cut points are removed through the inspection of the cut point value of each attribute to discretize the condition attributes and decision attributes in an information system. Experiment results show that the method is simple and effective.