Beyond Rebalancing: Benchmarking Binary Classifiers Under Class Imbalance Without Rebalancing Techniques
A new study published on arXiv explores the performance of binary classifiers when faced with imbalanced datasets, focusing on scenarios without rebalancing techniques. Researchers evaluated various classifiers, including traditional models and advanced ones like TabPFN and boosting ensembles, across real-world and synthetic datasets with progressively smaller minority class sizes. The findings indicate that classification difficulty increases with data complexity and reduced minority class representation, but advanced models demonstrate superior robustness and generalization compared to traditional methods. AI
IMPACT Provides guidance on model selection for imbalanced learning scenarios without relying on explicit rebalancing techniques.