Table 2 Summary of the benchmarking procedure for each dataset employed in this study
From: Tuning gradient boosting for imbalanced bioassay modelling with custom loss functions
Name | Split | Replicates | Metrics for external comparison | External baselines |
|---|---|---|---|---|
HIV | Random | 50 | ROC-AUC | RF, SVM, XGB, DNN, GCN, GAT, MPNN, AFP |
Tox21 | Random | 50 | ROC-AUC | RF, SVM, XGB, DNN, GCN, GAT, MPNN, AFP |
MUV | Random | 50 | PR-AUC | RF, SVM, XGB, DNN, GCN, GAT, MPNN, AFP |
Phosphatase | Scaffold | 5 | Accuracy, precision, recall, F1 score, ROC-AUC | DNN, GCN |
NTPase | Scaffold | 5 | Accuracy, precision, recall, F1 score, ROC-AUC | DNN, GCN |
HTS | Scaffold | 5 | Not applicable | Not applicable |