基于可辨识矩阵的启发式属性约简方法及其应用

Heuristic Attribute Reduction Based on Discernibility Matrix and Its Application

在基于可辨识矩阵的属性约简算法的基础上，提出了基于可辨识矩阵的计算属性重要性的方法，并以此作为启发式知识来约简决策表中的冗余属性。这种方法直接源于评审数据，思路清晰，拟合结果表明本约简算法合理、可靠。

For the purpose of making up the faultiness of attribute reduction algorithm based on discernibility matrix, the method of attribute importance based on discernibility matrix is put forward in this paper and is used as heuristic knowledge in deciding the sequence of attribute reduction. Compared with general methods, the result attained by this method rooted in evaluation data directly, so it has more reliability and strong persuasion.