A Heuristic Genetic Algorithm for Continuous Attribute Discretization in Rough Set Theory

Abstract

Continuous attribute discretization based on rough set is to got possibly minimum number of cuts, and at the same time it should not weaken the indiscernibility ability of the original decision system. In order to obtain the optimal cut set of the continuous attribute system, based on research the choice of candidate cut set, this paper presents a heuristic genetic algorithm for continuous attribute discretization to decision tables. In this algorithm making the importance of the continuous cut as heuristic message, a new operator is constructed to not only maintain the discernibility of the cuts selected, but also improve local search ability of the algorithm. Compared the performance of this method with the others’, this method is proved effective and superiority.