A Improved Method of Discretization of Continuous Attributes

Abstract Discretization of continuous attributes is one of the important steps in preprocessing of data analysis. In this paper, a new method of supervised discretization of continuous attributes based on entropy and hierarchical clustering guiding by level of consistency of decision table is introduced. This method makes use of the concept of the level of consistency of decision table in Rough Sets. According to the level of consistency of the produced decision table, the number of hierarchical cluster is adjusted dynamically in the first step. And then in the second step, we merge adjacent region based on entropy without damaging the level of consistency. Experiments show that this method is feasible.