基于粗集的频率约简与动态约简对不一致表分类的方法

摘要

According to the adverse effect of inconsistent decision tables on accuracy and stability of information classiifcation,this paper proposes a method of classiifcation for inconsistent decision tables by combining multiform frequency reducts with dynamic reducts based on rough set. The method firstly converts inconsistent decision tables into consistent decision tables by frequency reducts, then classiifes consistent decision tables by dynamic reducts. Experiment show:On the premise of avoiding the information loss, the method make the best of excellent characteristics of two reducts to improve accuracy and stability of information classification of inconsistent decision tables.%针对不一致决策表对于信息分类精度与稳定性的不利影响，提出基于粗集理论的利用不同形式频率约简与动态约简技术相结合的信息分类方法。首先利用频率约简对不一致决策表进行一致性处理，然后利用动态约简技术对获取的一致决策表进行分类。通过测试系统计算表明，充分结合两者的优良特性，在不损失数据信息前提下，提高了不一致决策表的分类精度和稳定性。