Table 2 Statistical error analysis for the models developed in this work.

From: Machine and deep learning models for predicting high pressure density of heterocyclic thiophenic compounds based on critical properties

Models

Statistical parameters

AAPRE

APRE

RMSE

SD

R2

LightGBM

 Train

0.02126

0.00040

0.32131

0.00031

0.99998

 Test

0.03034

− 0.00231

0.44628

0.00043

0.99997

 Total

0.02308

− 0.00014

0.34998

0.00033

0.99998

AdaBoost-DT

 Train

0.12346

0.00041

1.78314

0.00163

0.99961

 Test

0.18487

0.00811

2.61380

0.00240

0.99921

 Total

0.13578

0.00196

1.97794

0.00181

0.99953

GBoost

 Train

0.01967

0.00001

0.29777

0.00029

0.99998

 Test

0.05051

0.00212

0.70362

0.00067

0.99994

 Total

0.02585

0.00044

0.41254

0.00039

0.99997

DT

 Train

0.05648

− 0.00096

1.02467

0.00098

0.99987

 Test

0.25361

0.01116

3.03449

0.00282

0.99894

 Total

0.09602

0.00216

1.63904

0.00154

0.99968

TabNet

 Train

0.18276

− 0.06131

2.81115

0.00258

0.99905

 Test

0.17718

− 0.07547

2.64649

0.00245

0.99919

 Total

0.18164

− 0.06415

2.77890

0.00255

0.99908

DNN

 Train

0.09832

0.07863

1.44123

0.00125

0.99975

 Test

0.09970

0.07394

1.46719

0.00127

0.99975

 Total

0.09860

0.07769

1.44647

0.00126

0.99975

  1. The best results are in bold.
Back to article page