The Methods Of Classification Rule Extraction Based On Rough Sets Technique

The key problem of inductive learning is to extract classification rules from a givendataset, the induction of decision is a typical approach of of classification rule extraction. It isa crucial problem to select expanded attributes in the induction of decision tree. Theuncertainty of each cut of each continuous-valued attributes is needed to measure during theselection of expanded attributes for induction of decision tree based on discretion method, andthe computational time complexity is very high. In order to deal with this problem, a methodof rules extraction with decision tree for continuous-valued attributes based on tolerancerough sets technique is proposed in this paper. The method consists of three stages. Firstexpanded attributes are selected with tolerance rough sets technique, and then the optimal cutof the expanded attribute is found, the sample set is partitioned by the optimal cut and finallythe decision tree can be generated recursively. In addition, In order to deal with the problemof high complexity of the CNN (Condensed Nearest Neighbor), Especially, for a largedatabase, a CNN algorithm based on rough set technique is proposed in this paper, whichconsists of three stages: Firstly, to remove the superfluous attributes an attribute reduct iscomputed with rough set method. Secondly, the instances within boundary regions areselected. Meanwhile the redundant instances are removed. Finally, a CS is found from theselected instances. We analyze the computational time complexity of the algorithms proposedin this paper in theory and conducted some experiments on multiple database. Theexperimental results and the statistical analysis of the results demonstrate that the proposedmethods outperform other related methods in terms of computational complexity andclassification accuracy.