The class imbalance problem is a significant challenge in binary classification tasks, often leading to biased model predictions. This paper presents ‘RbImbD,’ a novel region-based method for addressing both class imbalance and data overlapping in a single framework. The proposed approach is evaluated on 66 synthetic datasets using metrics such as sensitivity, precision, and F1-score. Results show improved performance over existing techniques. While the method demonstrates strong results on synthetic datasets, its computational complexity can increase when applied to large, real-world datasets. Future work will focus on optimizing computational efficiency and extending the method to high-dimensional data. This approach uniquely integrates the handling of imbalance and overlap, setting it apart from traditional methods.