Table 3.
Parameters of selected machine learning algorithms.
| Model | Description |
|---|---|
| DecisionTreeClassifier (DT) | Standard decision tree regressor with max_depth = 10 |
| KNeighborsClassifier (KNN) | Standard classifier based on k-nearest neighbors |
| RandomForestClassifier (RF) | Standard random forest with n_estimators = 20, max_depth = 10 |
| Simple neural network (NN) | Multilayer neural network with 4 hidden Dense layers of 200 neurons with ReLU activation function, 1 Dropout layer (20% dropout rate) |
| Long short-term memory neural network (LSTM) | Multilayer neural network with two LSTM layers (20 and 50 neurons), 2 hidden Dense layers of 100 neurons, 1 Dropout layer (20% dropout rate) |
| Multiple neural network (CNN) | Multilayer Neural Network with 4 Blocks from the Conv1D Convolutional Layer (number of filters from 32 to 256, convolutional kernel = 3) combined with BatchNormalization, followed by GlobalAvgPool1D and 1 Dense layer of 100 neurons |
| CNN + Transformer (Transformer) | A model based on the MobileViT architecture shown in Figure 2 |