SMOTE-FRST: A NEW RESAMPLING METHOD USING FUZZY ROUGH SET THEORY
Enislay Ramentol, Nele Verbiest, Rafael Bello, Y. CABALLERO, Chris Cornelis, Francisco Herrera · World Scientific proceedings series on computer engineering and information science · 2012
In this paper, we introduce a new hybrid preprocessing method for editing imbalanced data. The algorithm we propose first resamples the training data using the Synthetic Minority Oversampling Technique (SMOTE) method, and subsequently applies an editing technique based on fuzzy rough set theory to the balanced training set. We evaluate the performance of our algorithm in an experimental study, using the C4.5 classifier as the learning algorithm. Statistical tests show the superiority of our method over state-of-the-art resampling methods.