Abstract: Real-world applications frequently use data with an unbalanced class distribution, meaning that the bulk of the data belongs to the majority class and the minority class is underrepresented. The classifier tends to anticipate that the majority of the incoming data will belong to the majority class in this scenario if all the data are utilized as training data. In the imbalanced class distribution problem, it is crucial to choose the appropriate training data for prediction and classification. In our project, we provide a unique hybrid algorithm with a mix of sampling strategies for choosing representative data as training data to enhance the prediction accuracy of dependent and independent data on an unbalanced class distribution problem.
Keywords: Imbalanced Data analysis, Oversampling, Under-sampling, Hybrid Sampling
|
DOI:
10.17148/IJARCCE.2024.134227
