一种基于属性重要度的约简算法

An Algorithm for Attribute Reduction Based on Attribute Significance

属性约简是Rough Set理论的核心内容之一,现已证明求决策表所有约简和最小约简是一个典型的NP难题。在对启发式约简算法研究的基础上,提出一种改进的约简算法。该算法将决策属性作为一个特殊的条件属性加入到核中,通过分析加入后划分出的等价类的变化来缩小需要处理数据的范围,并通过实例分析表明该算法的有效性。

The reduction of attributes is a core contents of Rough Set theory. It is proved that computing all the reduction and the optimal reduction of decision table is a NP typical problem. In this paper, heuristic algorithm is deeply investigated and an improved algorithm is proposed. An example shows the algorithm＇ s efficiency.