Fig. 4

Effects of the different synthetic resampling methods and the enrichment approach on the ML and DL algorithms’ performances based on three metrics: Balanced Accuracy, MCC and F1-score. a Models performances when trained on the different variations of the AID1063 dataset. b Models performances when trained on the different variations of theAID1258 dataset. c Models performances when trained on the different variations of theAID1706 dataset. d Models performances when trained on the different variations of theAID1479145 dataset