KNNOR: An oversampling technique for imbalanced datasets

Ashhadul Islam, Samir Brahim Belhaouari, Atiq ur Rehman, Halima Bensmail

Abstract: Highlights•A novel idea to characterize the compactness of imbalanced datasets.•Utilize each minority datapoint and its k-nearest neighbors while generating an artificial data point.•Ensure that noise and outliers can be ignored effectively.•Code has been made available as a python library.