Table 6 Performance evaluation of different ML models in CS prediction (testing phase).

From: AI driven prediction of early age compressive strength in ultra high performance fiber reinforced concrete

Ref.

Concrete type

(ML) models

RMSE

MAE

R2

Current study

UHPFRC

RF

10.863

7.882

0.816

GB

9.927

7.631

0.849

SVR

7.743

4.178

0.91

ANN

9.876

6.317

0.852

GPR

6.835

3.345

0.932

72

Normal concrete at early age

ANN

7.176

5.55

0.909

SVR

10.91

8.687

0.779

LR

10.97

8.634

0.768

ANN + SVR + LR

9.039

7.103

0.855

ANN + SVR + LR

8.36

6.622

0.883

ANN + LR

8.319

6.504

0.876

73

UHPC

CatBoost

6.4

4.88

0.96

PPSO-CatBoost

6.23

0.039

0.976

DMO-CatBoost

6.15

0.038

0.978

35

UHPC

MLP-ANN

11.734

8.187

0.77

DT

11.014

7.285

0.8

RF

9.335

6.485

0.85

SVR

9.67

6.971

0.84

KNN

10.63

7.67

0.81

BR

15.308

12.294

0.61

  1. Linear regression (LR), Particle Swarm Optimization-CatBoost (PPSO-CatBoost), and Differential Mutation Optimization-CatBoost (DMO-CatBoost), Multi-Layer Perceptron-ANN (MLP-ANN), Decision Tree (DT), RF, SVR, K-Nearest Neighbors (KNN), and Bayesian Ridge (BR).
Back to article page