基于粗糙集的属性约简算法

The attribute reduction algorithms based on rough sets

目的针对不完备信息系统寻找处理方法。方法利用粗糙集扩充模型限制容差关系,取代经典粗糙集理论中的不可分辨关系,对不完备信息进行处理。结果理论上证明了算法的完备性,并使用UCI机器学习数据库对属性约简算法进行测试。结论所述的属性约简算法,不仅可以处理不完备信息,而且在效率上有所提高。

Aim Find a feasible algorithm for incomplete information systems.Methods An extended rough set model is proposed to deal with the missing value which is called limited tolerance relation model to replace the non-distinguish relation.Results The completeness of the algorithm is theoretically proved,and the UCI machine learning database has been used for testing.Conclusion Test results show that the attribute reduction algorithm described in this paper,not only can deal with incomplete information,but also can improve the efficiency.