Table 2 Optimal hyperparameter values for each machine learning model.
Model | Key Hyperparameters | Optimal Values |
|---|---|---|
AdaBoost | Number of base estimators | 15 |
Learning rate | 1 | |
RF | Max depth | 21 |
Number of trees (n_estimators) | 100 (default) | |
Max features | sqrt (default) | |
DT | Max depth | 3.34 |
ANN | First hidden layer neurons | 25 |
Second hidden layer neurons | 17 | |
Transfer functions | Hyperbolic Tangent Sigmoid (tansig-hidden layers), Linear activation function (purelin- output layer) | |
LSSVM | Kernel | RBF |
Regularization parameter (γ) | 1213 | |
Kernel width (σ²) | 0.77 | |
1D-CNN | Convolutional layers | 2 |
Pooling layers | 1 | |
Fully connected layers | 1 | |
Activation function | Rectified Linear Unit (ReLU) | |
Learning rate | 0.001 | |
Epochs | 100 (default) | |
Batch size | 16 (assumed) | |
EL | Combined models | SVM, DT, KNN with C = 120, gamma = 0.02, epsilon = 0.001 |
SVM settings | RBF used as the Kernel function, distance = Euclidean | |
KNN settings | k = 9, distance = Euclidean | |
Combination method | Averaging |
