An optimized parallel decision tree model based on rough set theory

Abstract

This paper presents an optimized parallel decision tree model based on rough set theory, first the model divides global database into subsets, then using the intuitive classification ability of decision tree to learn the rules in each subset, at last merge each subset's rule set to obtain the global rule set. In this model, with the uncertain information analysis method of rough set, the author presents a massive data segmentation method and using the Weighted Mean Roughness as a decision tree method for attribute selection. This parallel data mining model with the best segmentation algorithm based on rough set can be well used in dealing with massive database © 2008 Springer-Verlag Berlin Heidelberg.