基于邻域关系的决策表约简

Decision table reduction based on neighborhood relation

针对经典粗糙集理论难以处理连续型数据的特点,提出基于邻域关系的决策表约简方法。该方法在连续型数据的决策表中引入邻域关系,通过邻域关系进行信息粒化,避免离散化过程带来的信息损失。通过定义邻域正域和邻域约简概念,分析邻域正域的单调性原理,提出基于邻域关系的属性重要度概念,进一步设计了两种启发式约简算法。理论分析与实例表明该方法是有效可行的。

In view of the fact that the classical rough set theory has difficulty dealing with continuous data, a reduction method was proposed based on neighborhood relation in the decision table. By the definitions of neighborhood relation and neighborhood parameter, each object in the universe was assigned to a neighborhood subset, called neighborhood granule, which could avoid the loss of information in the discretization process. The concepts of neighborhood positive region and neighborhood reduction were defined. The positive region monotonous principle was analyzed. Furthermore, the dependency function based on neighborhood relation was used to evaluate the significance of attributes and two heu- ristic attribute reduction algorithms were constructed. Theoretical analysis and an example showed that the reduction method was efficient and feasible.