Table 5 Predictive performance of the three regression models (RF, GB, and ET) based on global statistical metrics (MAE, RMSE, and R²).

From: Machine learning for quantitative LIBS analysis of aluminum alloys: a comparison of random forest, gradient boosting, and extremely randomized trees

Sample

Metric

RF model

GB model

ET model

Value

cp-Al-67990

MAE (%wt)

0.0294

0.1325

0.0125

RMSE (%wt)

0.0402

0.1664

0.0163

1.000000

0.999985

1.000000

cp-Al-11697

MAE (%wt)

0.1023

0.2483

0.1412

RMSE (%wt)

0.1422

0.4171

0.1914

0.999999

0.999986

0.999999

cp-Al-11628

MAE (%wt)

0.0876

0.1276

0.0501

RMSE (%wt)

0.1199

0.1767

0.0786

0.999999

0.999994

0.999998

Al-Cu_E113

MAE (%wt)

0.0556

0.2470

0.0506

RMSE (%wt)

0.0750

0.3095

0.0679

0.999998

0.999967

0.999998

Al-Cu_E115

MAE (%wt)

0.0934

0.2142

0.1885

RMSE (%wt)

0.1160

0.2779

0.2168

0.999989

0.999976

0.999992

Al-Cu_E116

MAE (%wt)

0.2036

0.2268

0.1519

RMSE (%wt)

0.2747

0.3246

0.2058

0.999951

0.999983

0.999973

Al-Zn_G77J5

MAE (%wt)

0.065715

0.115174

0.054853

RMSE (%wt)

0.073461

0.145719

0.064652

0.999996

0.999990

0.999999

Al-Cu-Zn_G77J1

MAE (%wt)

0.1224

0.1546

0.0940

RMSE (%wt)

0.1663

0.1881

0.1381

0.999975

0.999977

0.999983

Al-Cu-Zn_G77J2

MAE (%wt)

0.0383

0.0923

0.0414

RMSE (%wt)

0.0431

0.1076

0.0570

0.999999

0.999990

0.999997

Al-Cu-Zn_G77J6

MAE (%wt)

0.0612

0.0346

0.0169

RMSE (%wt)

0.0947

0.0410

0.0262

0.999994

0.999999

0.999999