To understand the Naive Bayes ML algorithms, a good start is understanding the **original Naive Bayes** probability function and what **conditional probability** implies.

Short answer :

In your training data, you can “learn” inferences such as “*what is the probability of seeing the word ‘money’ in spams*“. You can create a large number of those inferences between your dependent variables (your features) and the class you are trying to predict. Probabilities allow you to combine those inferences (multiplying the probabilities) to make predictions on new data the model has never seen before, allowing to class it in a certain class.

Note : I’ll add some links in the comments

Read more here: Source link