Diverse binary classification datasets

该数据集研究评估了来自不同领域的30个数据集上的类别不平衡处理技术。这些数据集被用于比较SMOTE、类别权重调整和决策阈值校准这三种方法相对于基线的有效性。性能评估采用了包括F1分数、F2分数、精确度、召回率、Brier分数、PR-AUC和AUC等多重指标。这一研究规模涉及30个数据集，任务集中在二分类问题上。

This dataset-based study evaluates class imbalance handling techniques across 30 datasets from diverse domains. These datasets are utilized to compare the effectiveness of three methods—SMOTE, class weight adjustment, and decision threshold calibration—against the baseline. Multiple performance metrics including F1-score, F2-score, precision, recall, Brier score, PR-AUC, and AUC are adopted for evaluation. This research involves 30 datasets and focuses on binary classification tasks.