midNaïve Bayes
What is log probability trick in Naïve Bayes and why is it important?
Updated May 17, 2026
Short answer
Log probabilities are used to prevent numerical underflow when multiplying many small probabilities.
Deep explanation
Naïve Bayes multiplies many probabilities, which can become extremely small and underflow floating-point precision. Taking logarithms converts multiplication into addition: log(P1 P2 ... Pn) = Σ log(Pi). This improves numerical stability.
Real-world example
Text classification with long documents containing thousands of words.
Common mistakes
- Using raw probabilities leading to underflow errors.
Follow-up questions
- Why is log addition equivalent?
- Does log affect classification result?