Hyperparameter tuning is essential for optimizing machine learning models. In this example, we’ll demonstrate how to use scikit-learn’s RandomizedSearchCV
for hyperparameter tuning of a SGDRegressor
model, commonly used for regression tasks.
Random search is a method for evaluating different combinations of model hyperparameters. Unlike grid search, it samples a fixed number of hyperparameter combinations from a specified distribution, making it more efficient when searching over a large hyperparameter space.
SGDRegressor
is a linear model that fits linear models using stochastic gradient descent (SGD). It is particularly suited for large-scale and sparse datasets. The model is trained by minimizing a loss function such as the squared loss for regression.
Key hyperparameters for SGDRegressor
include the learning rate (eta0
), which controls the step size in the parameter space during optimization; the penalty type (l1
, l2
, or elasticnet
), which determines the type of regularization applied to prevent overfitting; and the maximum number of iterations (max_iter
), which sets the number of passes over the training data.
from sklearn.datasets import make_regression
from sklearn.model_selection import train_test_split, RandomizedSearchCV
from sklearn.linear_model import SGDRegressor
from scipy.stats import uniform
# Generate synthetic regression dataset
X, y = make_regression(n_samples=1000, n_features=10, noise=0.1, random_state=42)
# Split into train and test sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Define the model
model = SGDRegressor(random_state=42)
# Define hyperparameter distribution
param_dist = {
'eta0': uniform(loc=0.001, scale=0.099),
'penalty': ['l1', 'l2', 'elasticnet'],
'max_iter': [100, 200, 500, 1000]
}
# Perform random search
random_search = RandomizedSearchCV(estimator=model,
param_distributions=param_dist,
n_iter=100,
cv=5,
scoring='neg_mean_squared_error',
random_state=42)
random_search.fit(X_train, y_train)
# Report best score and parameters
print(f"Best score: {random_search.best_score_:.3f}")
print(f"Best parameters: {random_search.best_params_}")
# Evaluate on test set
best_model = random_search.best_estimator_
mse = best_model.score(X_test, y_test)
print(f"Test set mean squared error: {mse:.3f}")
Running the example gives an output like:
Best score: -0.010
Best parameters: {'eta0': 0.014951498272501501, 'max_iter': 500, 'penalty': 'l1'}
Test set mean squared error: 1.000
The steps are as follows:
- Generate a synthetic regression dataset using scikit-learn’s
make_regression
function. - Split the dataset into train and test sets using
train_test_split
. - Define the model then the hyperparameter distribution with different values for
eta0
,penalty
, andmax_iter
hyperparameters. - Perform random search using
RandomizedSearchCV
, specifying theSGDRegressor
model, hyperparameter distribution, 100 iterations, 5-fold cross-validation, and mean squared error (MSE) scoring metric. - Report the best cross-validation score and best set of hyperparameters found by random search.
- Evaluate the best model on the hold-out test set and report the mean squared error.
By using RandomizedSearchCV
, we can efficiently explore different hyperparameter settings and find the combination that maximizes the model’s performance. This automated approach saves time and effort compared to manual hyperparameter tuning and helps ensure we select the best configuration for our SGDRegressor
model.