Smote-variants: A python implementation of 85 minority oversampling techniques

摘要

Imbalanced classification problems are definitely around He and Gracia (2009), and a successful approach to avoid the overfitting of majority classes is the synthetic generation of minority training samples Fernandez et al. (2018). Despite the large number of minority oversampling algorithms proposed, open source implementations are available for only a handful of techniques. The package smote-variants provides a Python implementation of 85 oversampling techniques to boost the applications and development in the field of imbalanced learning. The source code, documentation and examples are available in the GitHub repository http://github.com/gykovacs/smote_variants/. (C) 2019 Elsevier B.V. All rights reserved.