基于决策表的区分矩阵增量属性约简算法

Incremental attribute reduction algorithm based on discernibility matrix in decision table

对于决策表中存在对象动态变化的现象,当利用静态的属性约简算法处理这类决策表时算法效率并不理想,为了有效提高增量属性约简算法的效率,对决策表进行了简化,并证明了基于简化区分矩阵的属性约简与基于区分矩阵的属性约简是一致的,在利用原的属性约简的基础上,提出了一种基于决策表的区分矩阵增量属性约简算法,通过实例分析说明算法的有效性和可行性。

At present, many static algorithms of knowledge reduction based on decision table have been proposed, however, since the objects in the actual decision table are often changed, these static algorithms are computationally time-consuming. Aiming at the problem, the concept of simplified decision table is introduced, and it is proved that attribute reduction based on the simplified discernibility matrix is equivalent to that based on discernibility matrix. On this condition, an efficient incremental computing algorithm for attribute reduction based on decision table is designed. Example results illustrate the efficiency and feasibility of the new algorithm.