Abstract: The positive region reduction and variable precision reduction for decision table are two important types of attribute reductions. By comparing the calculation process of the two discernibility matrices, under the condition that the variable precision threshold is greater than 0.5, the value of the decision attribute can be appropriately changed, so that the calculation of the variable precision reduction can be converted into the calculation of the positive region reduction. This paper shows that the calculation of any variable precision reduction can be transformed into that of a positive region reduction. This transformation can improve the efficiency of the variable precision reduction algorithm.

Key words: variable precision reduction, positive region reduction, discernibility matrix, decision table, rough set

摘要： 正区域约简与变精度约简是决策表属性约简的两种重要类型，基于辨析矩阵的约简方法能够得到所有的约简结果，通过比较两者辨析矩阵的计算过程，在阈值大于0.5的条件下，可以适当改变决策属性的值，使得变精度约简的计算可以转化为正区域约简的计算，从而为变精度约简的计算提供了一种新的计算方法。举例说明，通过上述转换可以简化变精度约简的计算。

关键词: 变精度约简, 正区域约简, 辨析矩阵, 决策表, 粗糙集