Spaces:
Sleeping
Sleeping
File size: 1,014 Bytes
7c045bd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
"""
MLP Classifier setup.
Features:
- Uses `MLPClassifier`.
- Suitable for binary and multi-class classification.
- Default scoring: 'accuracy'.
Considerations:
- `hidden_layer_sizes`, `alpha` (L2 regularization), and `learning_rate_init` are common parameters.
- Increase `max_iter` if convergence warnings appear.
"""
from sklearn.neural_network import MLPClassifier
# Define the estimator
estimator = MLPClassifier(max_iter=200, random_state=42)
# Define the hyperparameter grid
param_grid = {
'model__hidden_layer_sizes': [(50,)], # Reduced size of hidden layers for faster training
'model__alpha': [0.001], # Retained commonly effective value
'model__learning_rate_init': [0.001], # Focused on a single typical value for faster tuning
# Uncomment and customize preprocessing params if needed
#'preprocessor__num__imputer__strategy': ['mean'],
#'preprocessor__num__scaler__with_mean': [True],
#'preprocessor__num__scaler__with_std': [True],
}
default_scoring = 'accuracy'
|