Naive Bayes Model
- is a type of probabilistic graphical model
- is the simplest Bayesian Network models
- is a generative model
- is a high bias model
- is based on applying Bayes’ theorem with strong (naïve) independence assumptions between the features
Bayes Model vs Naive Bayes Model
Bayes Model |
Naive Bayes Model |
|---|---|
|
𝑿 = [𝑋1, …, 𝑋𝑛] are observed variables | |
|
based on Bayes’ theorem
|
assumption that 𝑋1 through 𝑋𝑛 are conditionally independent given 𝑌
|
|
directly estimates parameters for:
|
directly estimates parameters for:
|
|
generative linear classifier - learns the joint probability distribution 𝐏(𝑿,𝑌) = 𝐏(𝑿|𝑌)𝐏(𝑌) |
generative linear classifier - learns the APPROXIMATE joint probability distribution 𝐏(𝑿,𝑌) ≈ 𝐏(𝑌) 𝛱1≤𝑖≤𝑛𝐏(𝑋𝑖|𝑌) |
|
|
|
/naive-bayes-model---bayes-model/logistic-regression-vs-naive-bayes-3.png)
/naive-bayes-model---bayes-model/logistic-regression-vs-naive-bayes-1.png)
Naive Bayes Model - Versions
- Gaussian Naive Bayes
- Multinomial Naive Bayes