A novel oversampling and feature selection hybrid algorithm for imbalanced data classification

Traditional approaches tend to cause classier bias in the imbalanced data set, resulting poor classification performance for minority classes. In particular, there are many financial fraud, network intrusion, and fault detection, where recognition rate of classes is pertinent than majority Therefore, pressure on developing efficient algorithms solve class imbalance problem. To this end, article presents a novel hybrid algorithm Negative Binary General (NBG), improve classifications by combining oversampling feature selection algorithm. A algorithm, Negative-positive Synthetic Minority Oversampling Technique (NPSMOTE), improves sample generation’s practicability while Ant Lion Optimizer (BALO) extracts most significant features performance. Simulation experiments carried out using seven benchmark sets demonstrate that, proposed NBG significantly outperforms small-sample compared nine other existing six recently published algorithms.