Comparison of binary classification methods in case of imbalance class

This study investigates the binary classification problem in the case of imbalanced class distribution. In case of imbalanced class distribution, sampling methods are utilized. In this way, a balanced class distribution is obtained. SMOTE-NC algorithm is used for this purpose. The performances of logistic regression, support vector machines and gradient boosting models on balanced data sets obtained under SMOTE-NC algorithm are analyzed. According to the obtained results, the use of SMOTE-NC algorithm together with gradient boosting increases the binary classification performance in case of imbalanced class distribution. SMOTE-NC algorithm increases the correct classification rate of the minority class.