Naive Bayes Classifier
Naive Bayes applies Bayes' theorem with independence assumption.
Bayes Theorem
P(class|features) ∝ P(features|class) × P(class). Independence assumption: P(features|class) = ∏P(feature|class).
Types
Gaussian: continuous features, assumes normal distribution. Multinomial: text, bag-of-words. Bernoulli: binary features.
Advantages
Fast, works well with high dimensions, handles missing data. Independence assumption often violated but still effective.
Key Takeaways
- Naive Bayes applies Bayes theorem
- Independence assumption simplifies computation
- Works surprisingly well despite unrealistic assumption