Skip to main content

Table 3 Summary of the results for the datasets belonging to the MoleculeNet repository

From: Tuning gradient boosting for imbalanced bioassay modelling with custom loss functions

Name

Metric

WCE

FC

LA

EQ

LDAM

Best descriptor-based

Best graph-based

HIV

ROC-AUC

0.811 ± 0.02

0.831 ± 0.01

0.823 ± 0.03

0.809 ± 0.02

0.833 ± 0.02

0.822 ± 0.02

0.833 ± 0.02

Tox21

ROC-AUC

0.790 ± 0.01

0.808 ± 0.01

0.812 ± 0.01

0.781 ± 0.02

0.808 ± 0.01

0.838 ± 0.01

0.852 ± 0.01

MUV

PR-AUC

0.152 ± 0.03

0.127 ± 0.02

0.140 ± 0.03

0.126 ± 0.03

0.141 ± 0.03

0.112 ± 0.04

0.061 ± 0.03

  1. The best values for each metric in each dataset are highlighted in bold