基于自组织特征映射的属性离散化方法

A Method of the Discretization of Rough Set's Continuous Attribute Based on the Self-organizing Feature Map

基于粗糙集理论,用K-W方法初步评价各连续属性的重要性,用自组织特征映射的聚类方法进行离散化,以决策表相容性为判决标准.试验表明,这是一种令人满意的离散化方法.

In this paper, we estimate the importance of continuous attributes with the method of K-W clustering method, and discrete continuous attributes using self-organizing feature map （SOFM）. This method uses the decision table＇s consistency as its discretization criterion. Experimentation indicates that it is a satisfaction method.