Table 3 Parameter optimization on the training dataset using fivefold cross-validation.
From: Prediction of concrete compressive strength using a Deepforest-based model
Regressor | Parameter range | Optimized parameters |
|---|---|---|
Linear | – | – |
KNeighbors | n_neighbors:[0, 100] | 4 |
DecisionTree | max_depth:[0, 100] | 36 |
SVR | C:[0, 10], gamma:[0, 10] | C:9.8, gamma:0.3 |
LASSO | alpha:np.logspace(− 4, 0, 50) | 0.13 |
MLP | hidden_layer_sizes:(10,10) ~ (100,100) | (50, 50) |
ExtraTrees | n_estimators:[200, 400] | 330 |
RandomForest | n_estimators:[200, 400] | 200 |
AdaBoost | n_estimators:[200, 400] | 240 |
GradientBoosting | n_estimators:[200, 400] | 100 |
Bagging | n_estimators:[200, 400] | 290 |
Deepforest | layers:[1, \(+\infty\)] | 2 |
