一种基于粗糙集理论的最简决策规则挖掘算法

研究粗糙集理论中可辨识矩阵，扩展了类别特征矩阵，提出一种基于粗糙集理论的最筒决策规则算法．该算法根据决策属性将原始决策表分成若干个等价子决策表．借助核属性和属性频率函数对各类别特征矩阵挖掘出最简决策规则．与可辨识矩阵相比，采用类别特征矩阵可有效减少存储空间和时间复杂度。增强规则的泛化能力．实验结果表明，采用所提出的算法获得的规则更为简洁和高效．

A mining algorithm for concise decision rules based on rough sets theory

By the research of discernibility matrix in rough sets, the extended class feature matrices are presented. A mining algorithm for concise decision rules based on rough set theory is proposed. The decision table is divided into many equivalence decision tables by using the decision attributes, core attributes are extracted and the attributes frequent functions are computed to mine decision rules from the small class feature matrices. Compared with algorithms based on discernibility matrices, the proposed algorithm is of much less space complexity and time complexity and has more generalizing ability. The experiment results on data sets in Rosetta software and comparison show that the algorithm provides more precise and simple decision rules.