Skip to main content

Table 2 Summary of 3D model performance

From: Use of structure-activity landscape index curves and curve integrals to evaluate the performance of multiple machine learning prediction models

Model

ANNE

SVM

MLR

KPLS

RF

PLS

Training set

      

MAE

0.19

0.21

0.20

0.22

0.10

0.22

Kendall

0.64

0.62

0.62

0.58

0.86

0.60

SCI

0.67

0.73

0.87

0.86

0.99

0.13

S(0)

0.64

0.62

0.62

0.58

0.86

0.59

S(1)

1.00

1.00

1.00

1.00

1.00

-1.00

Test Set

      

MAE

0.20

0.22

0.23

0.25

0.20

0.23

Kendall

0.55

0.52

0.49

0.43

0.56

0.48

SCI

0.93

0.83

0.83

-0.74

-0.66

0.73

S(0)

0.56

0.52

0.50

0.44

0.57

0.49

S(1)

1.00

1.00

-1.00

-1.00

-1.00

1.00

Prospective Set

      

MAE

0.35

0.34

0.74

0.34

*

0.32

Kendall

0.34

0.35

-0.09

0.38

*

0.34

SCI

-0.65

-0.49

-0.90

0.80

*

-0.69

S(0)

0.35

0.36

-0.07

0.39

*

0.35

S(1)

-1.00

-1.00

-1.00

1.00

*

-1.00

  1. Models using ADMET 3D Predictor descriptors and Kohonen map: ANNE, ADMET Predictor neural net; SVM, ADMET Predictor support vector machine; MLR, ADMET Predictor multiple linear regression; KPLS, ADMET Predictor kernel partial least squares; RF, Pipeline Pilot random forest; PLS, SIMCA-P+ partial least squares. The performance properties of the models were calculated as described in CALCULATIONS AND STATISTICS. The properties were not calculated for RF since prediction outliers could not be identified