Naive Bayes

Naive Bayes applies Bayes' theorem with the assumption that all features are conditionally independent given the class. Despite this simplification, it performs well for text classification and high-dimensional sparse data.

When to use:

• Text classification (spam detection, sentiment analysis) with bag-of-words features
• Very fast inference on high-dimensional data
• When probabilistic class estimates are needed and features are approximately independent

Input: Tabular data with the feature columns defined during training Output: Predicted class label and class probabilities

Model Settings (set during training, used at inference)

Variant (default: GaussianNB) The Naive Bayes variant determines the likelihood model:

• GaussianNB — assumes continuous features follow a Gaussian distribution
• MultinomialNB — for discrete count features (e.g., word counts)
• BernoulliNB — for binary features

Var Smoothing (GaussianNB, default: 1e-9) Adds a small value to variances for numerical stability.

Alpha (MultinomialNB/BernoulliNB, default: 1.0) Laplace smoothing parameter. Higher values apply stronger smoothing to unseen features.

Inference Settings

No dedicated inference-time settings. Class likelihoods are computed from the trained parameters.