论文部分内容阅读
针对目前基于决策表的属性约简矩阵算法效率不理想的情况,首先引入了简化决策表的概念,有效剔除了决策表中大量重复冗余的对象,并给出了正区域模型下简化矩阵的构造,从而有效地缩小了求解属性约简的搜索空间;然后从理论上详细分析了基于简化矩阵的属性约简与基于正区域的属性约简是一致的,并给出了一种有效的属性重要性度量方法.为此,在简化矩阵的基础上设出了一种快速的属性约简矩阵算法,最后通过实例分析和实验对比说明了算法的有效性和可行性.
Aiming at the current situation that the efficiency of attribute reduction matrix algorithm based on decision table is not ideal, the concept of simplified decision table is introduced at first, which effectively excludes a large number of redundant objects in decision table and gives the simplified matrix Structure, so as to effectively reduce the search space for solving attribute reduction; and then theoretically analyzed in detail the attribute reduction based on the simplified matrix is consistent with the attribute reduction based on the positive region, and gives an effective attribute Importance measure method.Therefore, a fast algorithm of attribute reduction matrix is proposed based on the simplified matrix. Finally, the effectiveness and feasibility of the algorithm are illustrated through the example analysis and experimental comparison.