Skip to main content

Table 4 Summary of the benchmarking results for the datasets in the MolData repository

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

Name

Metric

WCE

FC

LA

EQ

LDAM

DNN—Arshadi

GCN –Arshadi

Phosphatase

Accuracy

0.989 ± 0.0005

0.992 ± 4E-4

0.992 ± 3E-4

0.992 ± 7E-4

0.992 ± 2E-4

0.885

0.984

Precision

0.356 ± 0.01

0.455 ± 0.05

0.431 ± 0.06

0.571 ± 0.01

0.567 ± 0.05

0.027

0.144

Recall

0.139 ± 0.006

0.125 ± 0.01

0.135 ± 0.01

0.085 ± 0.02

0.109 ± 0.03

0.459

0.191

F1 score

0.200 ± 0.003

0.196 ± 0.01

0.206 ± 0.01

0.148 ± 0.01

0.182 ± 0.02

0.052

0.164

ROC-AUC

0.814 ± 0.0005

0.830 ± 0.001

0.830 ± 0.01

0.821 ± 0.0003

0.825 ± 0.0008

0.739

0.815

NTPase

Accuracy

0.945 ± 0.001

0.945 ± 0.004

0.945 ± 0.0004

0.899 ± 0.02

0.946 ± 0.005

0.854

0.933

Precision

0.381 ± 0.01

0.417 ± 0.01

0.472 ± 0.01

0.344 ± 0.04

0.488 ± 0.006

0.138

0.267

Recall

0.300 ± 0.007

0.294 ± 0.005

0.267 ± 0.003

0.250 ± 0.02

0.255 ± 0.005

0.526

0.095

F1 score

0.336 ± 0.003

0.345 ± 0.004

0.341 ± 0.005

0.289 ± 0.03

0.335 ± 0.003

0.219

0.141

ROC-AUC

0.821 ± 0.01

0.787 ± 0.01

0.852 ± 0.01

0.764 ± 0.007

0.827 ± 0.02

0.763

0.763

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