This table compares the performance of RF, BR, CC, and CSMLP algorithms across various datasets and features. The evaluation uses Multi-Label Random Under-Sampling (MLRUS) and Multi-Label Random Over-Sampling (MLROS) techniques with sampling ratios of 10%, 20%, and 30%. Micro-averaged F1-scores, Precision and Recall are reported for each algorithm-dataset pair; each cell has F1 score on the top, then Precision in the middle and Recall at the bottom. The last column shows percentage increases from the baseline compared to our best result to provide insights into handling class imbalance and improving classification accuracy. Additionally, the RF, BR, and CC results are compared with the findings from Saini <i>et al</i>. [8] on the IGD_FP dataset.

<p>This table compares the performance of RF, BR, CC, and CSMLP algorithms across various datasets and features. The evaluation uses Multi-Label Random Under-Sampling (MLRUS) and Multi-Label Random Over-Sampling (MLROS) techniques with sampling ratios of 10%, 20%, and 30%. Micro-averaged F1-sc...

Full description