A Novel Clustering Based Undersampling Algorithm for Imbalanced Data Sets Using Artificial Bee Colony Algorithm

Datasets that follow imbalanced class distribution impose a critical challenge to the machine learning research community. Standard machine learning approaches assume that datasets have a balanced class distribution and can’t handle imbalanced datasets. There are several other methods to deal with imbalanced datasets. This paper proposed an Artificial Bee Colony (ABC) clustering-based undersampling approach for imbalanced datasets. The proposed approach has been applied on several benchmark datasets from the KEEL dataset repository, and the obtained AUC value is compared and analyzed. We strongly recommend the Artificial Bee Colony-based undersampling method for dealing with imbalanced classification problems based on obtained results.