Research paper examining how data characteristics beyond imbalance ratio affect the effectiveness of oversampling methods in handling class imbalance. Proposes that method selection should account for broader dataset properties, not just imbalance severity.
Research
Beyond Imbalance Ratio: Data Characteristics as Critical Moderators of Oversampling Method Selection
Study reveals that effective oversampling method selection depends on broader data characteristics beyond imbalance ratio alone, reshaping how practitioners handle class imbalance.
Tuesday, April 7, 2026 12:00 PM UTC2 MIN READSOURCE: arXiv CS.LG (Machine Learning)BY sys://pipeline
Tags
research