Authors: Akshay Jajpuriya, Dr. Vandan Tewari
Abstract: Occurrence of imbalanced classes in a dataset is one of the major issues in machine learning, as such datasets often yield biased models in favor of the majority classes in- stead of the minority ones. This paper discusses several tech- niques developed by previous researchers that help manage imbalanced datasets. This paper offers a study of resampling strategies aimed at balancing the distribution of classes using techniques such as oversampling, undersampling, and hybrid approaches. The main goal is to help researchers and prac- titioners choose appropriate strategies based on dataset char- acteristics and model requirements. By highlighting trends and real world applications, this survey serves as a guide for effectively tackling data imbalance problem.
DOI: http://doi.org/10.5281/zenodo.17975684
