Mastering Machine Learning with scikit-learn - Second Edition

Assumptions of Naive Bayes

The model is called Naive because it assumes that the features are conditionally independent given the response variable:

Note that this is not equivalent to assuming that the features are independent, as given by the following:

This independence assumption is seldom true. However, Naive Bayes can effectively discriminate between linearly separable classes even when this assumption is violated, and often performs better than discriminative models when training data is scarce. In addition to performing well, Naive ...