Using a Novel Merit for Feature Selection Based on Rough Set Theory

Feature selection in microarray datasets has become one of the most interesting subjects in machine learning and data mining. Binary microarray datasets often consist of thousands of features with a small number of samples and the distribution of classes in them is imbalanced. This paper proposes a new merit based on rough set theory that is inspired by correlation-based merit and named rough set-based merit. Rough set-based merit is applied in rough set quick reduct algorithm to select a significant subset of features. The experimental results in section 4, show the robustness and benefits of the proposed method versus other existing methods in the literature.