Table 3 Comparison of statistical metrics for evaluating the performance of the GPR, ANN, and GEP models.

From: Predicting California bearing ratio of HARHA-treated expansive soils using Gaussian process regression

Model

Data set

R2

MAE

RMSE

RRMSE

ρ

OF

GPR (this study)

Training

0.9999

0.0920

0.13907

0.0078

0.00391

0.003

Testing

0.9997

0.2099

0.51819

0.0155

0.00775

Validation

0.9996

0.0719

0.1070

0.0025

0.00125

ANN29

Training

0.9998

0.0962

4.98

0.20

0.100

0.077

Testing

0.9997

0.2198

4.76

0.20

0.104

Validation

0.9994

0.1649

1.19

0.05

0.028

GEP30

Training

0.9972

0.5

4.94

0.202

0.101

0.028

Testing

0.9916

0.3

3.69

0.271

0.136

Validation

0.9932

0.5

5.49

0.167

0.084

Back to article page