粗糙模糊决策表概率决策分析的扩展粗糙集方法

Extension of rough sets model for probabilistic decision analysis from rough-fuzzy decision tables

应用λ-截集将决策类中的模糊集合转换为普通集合,在此基础上推广了粗糙隶属函数,讨论了其中的一些集合理论性质,通过设定置信阈值参数α,提出了一种可以从粗糙模糊决策表中获取概率决策规则的扩展粗糙集方法,并设计了一种改进的快速约简算法,最后给出了该方法的一个算例.研究结果表明,提出的方法可从冗余的且有噪声的粗糙模糊决策表中获取用于指导实践的概率决策知识.

The fuzzy set in decision class is transformed into ordinary set by applying a λ-cut, based on which the rough membership function is generalized and some set theoretic properties are discussed. After a confident threshold value is set, an extension of rough sets model is proposed which can obtain probabilistic decision rules from rough-fuzzy decision tables. An improved quick reduct algorithm is also presented. Finally, the approach is illustrated by an example. The results show that the proposed approach can extract probabilistic decision knowledge to guide practice from roughfuzzy decision tables with redundant noisy data.