Abstract
The problem of class imbalance has become a relevant topic in the field of machine learning, and there are several ways to solve this problem, including oversampling, undersampling, and hybrid algorithms. The HRBwSD algorithm proposed in the research belongs to the category of hybrid algorithms. The algorithm was created based on the RBU (Radial-Based Undersampling) and RBO (Radial-Based Oversampling) algorithms, which take into account the exclusion of duplicates and the preservation of the original class distribution. The results show that the model estimates obtained using the proposed algorithm were comparable to the estimates obtained using other class balancing algorithms. This confirms the importance of adapting algorithms to the specifics of the data to obtain more accurate and fair results.