Based on probability theory, naive Bayes is one of the simplest classification algorithms. If used properly, it can come up with accurate predictions. The Naive Bayes Algorithm is s0-named for two reasons:
- It is based on a naive assumption that there is independence between the features and the input variable.
- It is based on Bayes, theorem.
This algorithm tries to classify instances based on the probabilities of the preceding attributes/instances, assuming complete attribute independence.
There are three types of events:
- Independent events do not affect the probability of another event occurring (for example, receiving an email offering you free entry to a tech event and a re-organization occurring in your company).
- Dependent events affect the probability of another event occurring; that is, they are linked in some way (for example, the probability of you getting to a conference on time could be affected by an airline staff strike or flights that may not run on time).
- Mutually exclusive events cannot occur simultaneously (for example, the probability of rolling a three and a six on a single dice roll is 0—these two outcomes are mutually exclusive).