I am confused about Naive Bayes. Some sources e.g. here suggest only linear decision boundaries can be fit. Others here on SE or here on CV say or just suspect they can be non-linear. What is true? Are both suppositions based on different assumptions, perhaps?

Specifically the latter two source says that they can be linear if distributions are taken from the exponential family but this makes naive Bayes not per-se linear.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.