基于“桶”分类的快速属性约简算法

粗糙集中的属性约简算法如PawlakZ约简算法及Skowron算法均可有效找到约简的属性集,但都要花费巨大的时间、空间,且对于大数据集约简几乎不可能。针对以上经典算法的不足,文章提出一种快速属性约简算法。算法先将所有样本按决策属性分类,且用不同颜色的桶来装样本。其次,通过深度搜索结合广度搜索的策略,利用几个条件属性集的交集划分得到数据集的约简。在运算过程中,不断地删除交集中的数据,可有效提高效率。实验表明,该算法切实可行,可快速地找到属性约简集,并结合实例,说明了算法的可行性。

Fast attribute reduction algorithm based on classification barrels

Attribute reduced algorithms of rough sets have Pawlak A algorithm and Skowron which can effectively find reduced attribute sets.But both spent a lot of time, space, and almost impossible to complete for large data sets.To deal with the lack of classical algorithms, the article presents a fast attribute reduction algorithm.At first, all samples will be classified according to the decision-making properties, and use different colors of barrels to load them.The second, though in-depth search combined with the...