An extensive analysis of several methods for classifying unbalanced datasets

Abstract

In large-scale data applications, handling unbalanced data is a major issue. In order to gather the uneven data at the fastest pace feasible, the imbalanced data categorization system was created. Numerous neural methods have been developed to accurately categorize unbalanced data. However, because of the intricacy of the data, the classification process becomes more challenging due to increased resource utilization, computing costs, and algorithm complexity. As a result, this research has provided specifics on the performances of many classification models in various unbalanced datasets. Ultimately, a performance study was conducted to evaluate each model's categorization performance. For this reason, the precision, specificity, accuracy, and sensitivity have been used to measure the robustness. Each model's advantages and disadvantages are also thoroughly covered. The categorization models then offered future approaches to enhance the unbalanced data based on the drawbacks.