Attribute Reduction Algorithm and the Generation of Hybrid Decision Tree Based on Discernibility Matrix in Rough Set

Abstract

Through the use of the Discernibility Matrix in Rough Set, this paper introduced an Attribute Reduction Algorithm, based on which, a new one is put forward about the Generation of Hybrid Decision Tree. This Algorithm improved the traditional method as the attributes with high frequency of occurrence in the Discernibility Matrix can classify more examples at a time. Finally, by the comparison between the Algorithm and ID3, the new Algorithm is proved to be more superior and advantageous.