粗糙集中属性约简的一个贪心算法

A Greedy Algorithm for Attribute Reduction in Rough Set

利用单属性的逼近精度、由决策属性定义划分的粗糙逼近精度以及它们的均值和方差 ,给出了属性重要性程度的一种度量方式。在此基础上 ,提出了粗糙集中属性约简的一个贪心算法 ,将各属性按照重要性由大到小依次加入到约简属性集中 ,直到满足约简条件为止 ,其特点是简单、容易实现 ,在条件属性较多的情况下 ,往往能够迅速求得一个属性约简。

In this paper, we present a strategy for measuring the significance of an attribute based on the approximation qualities for each attribute, rough approximation quality for the partition defined by decision attribute and their mean values and square of variances. By virtue of this strategy, a greedy algorithm for attribute reduction is proposed, which constantly adds attributes to the candidate reduction according to their significance arranged in the decreasing order until a reduction is reached. Being simple and easily implemented, the presented algorithm can quickly find a reduction when many attributes are involved.