Skip to main content

Table 2 ML nano-QSTR models performance for training, validation, and test sets

From: A Nano-QSTR model to predict nano-cytotoxicity: an approach using human lung cells data

Model

Subset

R2

R2ext

Q2LOO

Q2F1

Q2F2

RMSE

MAE

CCC

DT

Training

0.953

-

-

-

-

6.147

3.840

0.976

Validation

0.733

-

0.730

-

-

14.103

8.441

0.851

Test

-

0.765

-

0.765

0.765

13.264

7.139

0.868

RF

Training

0.965

-

-

-

-

5.299

2.963

0.981

Validation

0.768

-

0.767

-

-

13.266

7.908

0.866

Test

-

0.790

-

0.789

0.789

12.573

6.879

0.881

ET

Training

0.953

-

-

-

-

6.158

3.912

0.975

Validation

0.798

-

0.797

-

-

12.054

7.294

0.889

Test

-

0.788

-

0.788

0.788

12.580

6.603

0.883

  1. DT: decision tree; RF: random forest; ET: extra-trees regressor; R2: determination coefficient (R2ext for external validation); Q: determination coefficient based-metrics (Q2LOO for internal validation; Q2F1 and Q2F2 for external validation); RMSE: root-mean-square error; MAE: mean absolute error; CCC: coefficient of concordance.