Scikit-Learn GridSearchCV CategoricalNB

Hyperparameter tuning is a crucial step in optimizing machine learning models for best performance. In this example, we’ll demonstrate how to use scikit-learn’s GridSearchCV to perform hyperparameter tuning for CategoricalNB, a Naive Bayes classifier designed for categorical data.

Grid search is a method for evaluating different combinations of model hyperparameters to find the best performing configuration. It exhaustively searches through a specified parameter grid, trains and evaluates the model for each combination using cross-validation, and selects the hyperparameters that yield the best performance metric.

CategoricalNB is a variant of the Naive Bayes algorithm that works with categorical features. It uses conditional probability for classification based on Bayes’ theorem. This model is suitable for classification tasks where the features are categorical in nature.

The key hyperparameters for CategoricalNB include the smoothing parameter alpha, which helps to prevent zero probabilities by adding a small value to the probability estimates, and fit_prior, which determines whether to learn class prior probabilities from the data or use uniform priors.

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.naive_bayes import CategoricalNB
import numpy as np

# Generate synthetic categorical dataset
X, y = make_classification(n_samples=1000, n_features=10, n_informative=5, n_redundant=5, random_state=42)
X = np.random.randint(0, 3, size=X.shape)  # Convert features to categorical by random integers

# 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 parameter grid
param_grid = {
    'alpha': [0.1, 0.5, 1.0],
    'fit_prior': [True, False]
}

# Perform grid search
grid_search = GridSearchCV(estimator=CategoricalNB(),
                           param_grid=param_grid,
                           cv=5,
                           scoring='accuracy')
grid_search.fit(X_train, y_train)

# Report best score and parameters
print(f"Best score: {grid_search.best_score_:.3f}")
print(f"Best parameters: {grid_search.best_params_}")

# Evaluate on test set
best_model = grid_search.best_estimator_
accuracy = best_model.score(X_test, y_test)
print(f"Test set accuracy: {accuracy:.3f}")

Running the example gives an output like:

Best score: 0.557
Best parameters: {'alpha': 0.1, 'fit_prior': True}
Test set accuracy: 0.550

The steps are as follows:

1. Generate a synthetic dataset with categorical features using make_classification and convert features to integers.
2. Split the dataset into train and test sets using train_test_split.
3. Define a parameter grid with different values for alpha and fit_prior hyperparameters.
4. Perform grid search using GridSearchCV, specifying the CategoricalNB model, parameter grid, 5-fold cross-validation, and accuracy scoring metric.
5. Report the best cross-validation score and best set of hyperparameters found by the grid search.
6. Evaluate the best model on the hold-out test set and report the accuracy.

By using GridSearchCV, you can efficiently explore different hyperparameter settings to maximize the model’s performance for categorical data classification tasks. This approach automates the process of hyperparameter tuning and ensures the best configuration for CategoricalNB.