What are the assumptions of Naïve Bayes?

Updated May 17, 2026

Short answer

Naïve Bayes assumes feature independence and equal contribution of features within a class.

Deep explanation

The key assumption is conditional independence: features are independent given the class label. It also assumes feature contributions are multiplicative in likelihood computation. While unrealistic, this assumption simplifies computation and avoids complex covariance estimation.

Real-world example

Text classification where each word is treated independently.

Common mistakes

  • Trying to use Naïve Bayes for strongly correlated numerical features.

Follow-up questions

  • What happens if features are correlated?
  • How can we reduce correlation impact?

More Naïve Bayes interview questions

View all →