Table 1 Performance of RF classifiers under different classification criteria

From: Machine learning-assisted amidase-catalytic enantioselectivity prediction and rational design of variants for improving enantioselectivity

RF-classifiers

criterion (\(-\Delta \Delta {G}^{\ne }\) in kcal/mol)

ee (%)

number of selected features

accuracy

recall

precision

F-score

AUC

Classifier A

1.86

80

56

0.978a

0.993

0.976

0.984

0.997

0.784b

0.903

0.801

0.831

0.768

Classifier B

2.40

90

62

0.980a

0.990

0.977

0.983

0.998

0.790b

0.859

0.809

0.831

0.832

Classifier C

3.00

95

47

0.937a

0.936

0.910

0.922

0.987

0.751b

0.739

0.676

0.788

0.701

  1. aPerformance matrices of training set (similarly hereafter); b Performance matrices of test set (similarly hereafter). RF: Random Forest, AUC: the area under receiver operating characteristic curve.
Back to article page