连续值域决策表的一种属性权重确定方法

An Algorithm of Obtain Weight of Continuous Attributes in Continuous Domain Decision Table

结合模糊聚类技术与粗糙集中属性重要性思想,针对连续值域决策表,提出一种属性权重的确定方法。该方法首先将连续值域决策表转化为模糊决策表,然后求出对象之间关于条件属性的相似度矩阵,再采用平方法将相似度矩阵化为等价矩阵,最后根据模糊聚类技术与粗糙集中求属性重要性方法求各连续属性的权重。实例说明了该方法的合理性和有效性。

Combining the technology of fuzzy clustering and ideal of rough sets, we put forward a method to compute the weight of continuous attributes in decision table. This method includes these steps： 1） turn continuousdomain decision table into fuzzy decision table; 2） compute the similarity matrix of objects with respect to condition attributes; 3） turn this similarity matrix into equivalence matrix; 4） compute weight of each continuous attribute according to fuzzy clustering and rough sets. Example shows this algorithm is reasonable and available.