Table 6 Performance criteria for the models during the training and testing phases.

From: Comparative assessment of standalone and hybrid deep neural networks for modeling daily pan evaporation in a semi-arid environment

Models

Training

R2

RMSE

NRMSE

MAE

NSE

PBIAS

DNN

0.797

5.000

0.656

4.742

0.002

− 60.31

DNN-SVM

0.795

2.000

0.262

1.630

0.750

0.421

DNN-BART

0.789

2.000

0.262

1.682

0.788

0.388

DNN-RSS

0.809

2.000

0.262

1.583

0.809

− 0.157

DNN-M5 pruned

0.795

2.000

0.262

1.645

0.794

0.419

DNN-RF

0.924

1.000

0.131

0.958

0.922

1.010

Models

Testing

DNN

0.668

5.000

0.654

4.870

0.061

− 58.099

DNN-SVM

0.651

3.000

0.393

2.127

0.649

3.540

DNN-BART

0.630

3.000

0.393

2.220

0.630

0.761

DNN-RSS

0.631

3.000

0.393

2.165

0.628

1.744

DNN-M5 pruned

0.635

3.000

0.393

2.166

0.633

2.309

DNN-RF

0.638

3.000

0.393

2.155

0.638

0.411

Back to article page