This is a very simple algorithm. For a relatively small training dataset, if features are independent, Naïve Bayes will usually perform well. For a large dataset, Naïve Bayes will still work well as feature independence can be assumed in this case, regardless of the truth. The training of Naïve Bayes is usually faster than any other algorithms due to its computational simplicity. However, this may lead to a high bias (but a low variance).