What is Laplace smoothing in Naïve Bayes?

Updated May 17, 2026

Short answer

Laplace smoothing prevents zero probabilities by adding small constants to counts.

Deep explanation

If a feature never appears in training data for a class, probability becomes zero, breaking Naïve Bayes. Laplace smoothing adds 1 (or α) to all counts to avoid zero probabilities and improve generalization.

Real-world example

Handling unseen words in spam detection systems.

Common mistakes

  • Not applying smoothing leading to zero probability collapse.

Follow-up questions

  • What is alpha smoothing?
  • Why is zero probability dangerous?

More Naïve Bayes interview questions

View all →