一种改进的基于粗糙集理论的启发式特征选择算法

An Improved Heuristic Feature Selection Algorithm Based on Rough Set Theory

在最大支持启发函数的基础上提出了一种新的基于粗糙集的启发函数,称为参数加权平均支持启发函数.该方法的优点是考虑了可能性规则集的整体质量,它所选出的特征在决策类上能形成具有高加权平均支持度的规则,并且能够利用阈值调整下近似的水平.计算实例表明该方法是有效的.

This paper proposes a new rough set-based heuristic function called parameterized weighted average support heuristic （PWASH） which is based on maxmium support heuristic （MSH）. The main advantage is that it considers the overall quality of the set of potential rules. PWASH selects features with high weighted average support of rules over all decision classes, and uses threshold to adjust the level of the lower approximation. Finally,the example proves this method is valid.