JISARA

Abstract: In the era of big data, the prevalence of imbalanced datasets has emerged as a significant challenge in machine learning and data analytics. Analysts often employ two primary techniques - undersampling and oversampling - to overcome the imbalance problem. This study explores multiple oversampling techniques in addressing these imbalances, focusing on how appropriate sampling methods can enhance model performance, improve predictive accuracy, and facilitate better decision-making. The results affirm that oversampling can improve the predictive power for the minority class when compared to building a model with imbalanced data. However, the additional contribution is that the type of balancing technique matters to the overall performance and accuracy of the predictive model.

Download this article: JISARA - V18 N2 Page 52.pdf

Recommended Citation: Tourt, D., Booker, Q., Rebman Jr., C.M., Jin, S.S., (2025). A Comparative Analysis of Oversampling Methods for Predicting Credit Card Default with Logistic Regression. Journal of Information Systems Applied Research and Analytics 18(2) pp 52-63. https://doi.org/10.62273/GLAH4676