基于辨识矩阵的属性集重要度评价方法

在决策表中，为了评价某条件属性的重要性，不但要考虑这个属性（单一属性）相对于决策属性的重要性，还要考虑该条件属性与其他条件属性构成的属性集的重要性。本文在属性集依赖度比单一属性依赖度更加可信的事实基础上，提出了一个基于辨识矩阵的属性集重要度评价方法。该方法能够较快地获得分辨矩阵，并直接求出属性集的依赖度，从而大大降低了算法的时间复杂度。实例验证了该方法具有较好的有效性和较低的时间复杂度。

Importance Degree Evaluation Method on Attribute Set Based on Discernable Matrix

In decision table, in order to evaluate importance of one condition attribute, the importance of the condition attribute with respect to decision attribute and the importance of the attribute set which is composed of the condition attribute and others attributes must be considered simultaneously. According to the fact that dependency degree of attribute set is more authentic than dependency degree of single attribute, a new importance degree evaluation method on attribute set is proposed based on discernable matrix. The proposed method can quickly get discernable matrix and directly obtain dependency degree of attributes set, so that time complexity of the proposed method is lower. Examples show the proposed method has better effectiveness and lower time complexity.