决策表中属性的重排

Attribute derangement in decision tables

研究决策表中属性的重排。对同一个决策表,把其中不同的属性分别指定为结论属性,借助决策表中属性之间的约简和冗余关系判别哪种重排可以导致粗糙集以及哪种属性重排导致非粗糙集。提出了属性重排粗糙度的概念,基于属性重排粗糙度提出了属性最佳逻辑流向、最佳逻辑属性等概念。分析并指出不同属性重排所对应的粗糙度和属性最佳逻辑流向反映了决策表中不同属性之间的因果蕴含关系。数值算例表明,利用决策表的属性重排可以有效地对缺失数据进行插补。

Attribute derangement in decision tables was proposed. Based on the redundant relations between attributes, the criteria was given for judging which kind of derangement led to rough set and which one would not. The notion of derangement roughness was proposed and then the notion of best logic attribute and the best attribute logic flow were put forward. It was shown that the best attribute logic flow reflected the causal relations between different attributes in deci sion tables. Numerical experiment showed that missing data could be imputed effectively by using attribute derange ment.