An attribute reduction method based on core samples set

For attribute reduction of an inconsistent decision information system, we first analyze the role of core samples set in the process of attribute reduction, then we introduce the concept of β-core samples set. By defining the similarity metric between two sample sets, we further establish an attribute reduction method based on core samples set, and give the concrete implementation step. Finally we analyze the feature and performance of our method using Postoperative Patient data set and Wine data set. The results indicate that our method has good interpretability and operability. So it can perform the attribute reduction of an information system under different decision preference.