Table 2 Tested options selected for the optimization of the developed ML.
Method | Hyper-parameter | Available options | Optimum option |
|---|---|---|---|
RF | Maximum features | [“auto”, “sqrt”, “log2”] | sqrt |
Maximum depth | [3, 4, 5, …, 30] | 25 | |
N of estimators | [3, 4, 5, …, 150] | 125 | |
GBR | learning rate | 0.1–0.9 | 0.21 |
estimators | 3-150 | 50 | |
subsample | 0.1–0.9 | 0.5 | |
DT | adept | 2–20 | 9 |
max_features | [“auto”, “sqrt”, “log2”] | sqrt | |
ANN | Number of nets | 1–5 | 4 |
Number of Neurons | 5-128 | 64 | |
net | Sequential,… | Sequential | |
Activation Function | ‘relu’, ‘tanh’ | ‘relu’, ‘tanh’ | |
SVM | kernel | ‘rbf’, ‘poly’, ‘sigmoid’ | rbf |
C | 0.1-10000 | 5000 | |
gamma | ‘scale’, ‘auto’ | scale |
