Research Of Rules Extraction Method Based On VPRSM In Incomplete Information Systems

Now knowledge discovery faces dealing with incomplete and uncertain data inefficiently and the poor explanation of knowledge. Variable Precision Rough Set(VPRS) theory is a kind of mathematic tool for the analysis of vague and uncertain knowledge which is no requirements of previous knowledge, and decision tree can directly analysis data。Therefore, using VPRS theory for attribute reduction and construction of decision tree to obtain rules for knowledge discovery has the extremely vital significance.According to the low accuracy of attributes reduction in the incomplete information system, this paper defines a new attribute importance operator as heuristic information and a new reduction definition as reduction terminal conditions,and expand relatively nuclear calculation method to incomplete information system. Finally, this paper proposes an attribute reduction algorithm based on VPRSM in incomplete information systems. This algorithm guaranteed to get an attribute reduction set,and can improve the accuracy and the flexibility of the attribute reduction. Because in the process of building a decision tree has not considered the problems of attribute importance in the nodes choice, therefore how to combine attribute nuclear and classification quality to decision tree construction, and when two(or more) attributes classification quality is equal, how to select the most appropriate attributes as node, this paper proposes the construction of decision tree algorithm based on VPRSM of Attribute Core. This algorithm can generate decision tree with less tree notes,higher classification accuracy, thus finally obtain rules.Theoretical analysis and experimental results on UCI data set prove that attribute reduction algorithm based on VPRSM in incomplete information systems and construction of decision tree algorithm based on VPRSM of Attribute Core can achieve the expected results, and get effective knowledge rules. The algorithms have important significance for future studies and practical applications.