一种基于粗糙集的特征选取算法及应用

特征选取技术主要目的之一是选取代表问题域的最优特征子集,提高分类的有效性和可伸缩性。文章在粗糙集理论框架下,提出了一种能处理不相容数据的特征选取算法,它以属性重要性和平均规则支持度作为特征选取的启发式信息,并且通过实验验证该算法的有效性。最后,将该算法应用到客户满意度特征选取中,对于识别和保留客户具有重要现实意义。

A Feature Seletion Algorithm Based on Rough Set Theory and its Applications

One of the aims of feature selection techniques is to select optimal attribues that stand for discussed field, and improve the effectivity and flexibility of classification.A new heurist feature selection algorithm that can properly process data noises is proposed under the frame of rough set theory.The algorithm based on the principles of attribute significance and rule support,and the experimental results show this algorithm is effective.As a practice,the algorithm is applied in customer feature selection to recognize and retain customers.