Table 3 Prediction error indicators for ML models.

From: Explainable automl and multi-objective optimization for sustainable high-performance geopolymer concrete

Error indicator

Data sets

Model

RF

XGBoost

LightGBM

AutoML

R2

Training set

0.8577

0.9526

0.9406

0.9451

Testing set

0.7971

0.9096

0.8731

0.8791

Validation set

0.8497

0.9270

0.9260

0.9280

RMSE (MPa)

Training set

6.3193

3.6481

4.0839

3.9242

Testing set

7.6902

5.1338

6.0828

5.9378

Validation set

7.6494

5.3307

5.3671

5.2954

MAE (MPa)

Training set

4.8270

2.1522

2.5486

2.3796

Testing set

5.9949

3.3976

4.2228

4.0224

Validation set

6.1662

4.0911

4.2245

4.2307

MAPE

Training set

0.0915

0.0385

0.0504

0.0457

Testing set

0.1277

0.0741

0.0874

0.0858

Validation set

0.1098

0.0743

0.0763

0.0724

a20

Training set

0.9223

0.9355

0.9369

0.9660

Testing set

0.8103

0.9138

0.8793

0.8966

Validation set

0.8065

0.9355

0.9032

0.9677

Back to article page