How does Naïve Bayes behave under Bayesian prior misspecification?

Incorrect priors can bias Naïve Bayes predictions, especially in small datasets.

Naïve Bayes relies on prior P(C) to scale posterior probabilities. If priors are incorrectly specified (e.g., uniform when data is skewed), decision boundaries shift. However, as dataset size grows, likelihood dominates and reduces prior sensitivity. This reflects Bayesian consistency properties.