An Open Access Journal
Table 3 Hyperparameter ranges where grid search was performed
From: Models for forecasting the traffic flow within the city of Ljubljana
Model type | Hyperparameter | Values |
|---|---|---|
Neural network | Depth | {1,2,3,4,5} |
Layer size | {128, 256, 512} | |
Dropout | {0, 0.2} | |
Early stopping patience | {5} | |
Optimizer | {Adam} | |
Decision tree | Maximum depth | {2, 4, 6, 8, 10, None} |
Minimum samples in leaf | {1, 2, 3, 4, 5} | |
Extremely randomized trees | Maximum depth | {2, 4, 6, 8, 10, None} |
Minimum samples in leaf | {1, 2, 3, 4, 5} | |
Linear regression | Optimizer | {SGD} |