Classification ML models #

Accuracy score doesn’t give correct measure in case of class imbalance. It doesn’t tell anything about the underlying class distribution. Cannot figure the type of error it is making. False positive, false negative etc.

Probability generative models are based on P(X, Y) and discriminative models are based on P(X|Y)

Discriminative models learns soft and hard boundaries b/w classes Logistic Regression, Support Vector Machine and Conditional Random Fields are popular discriminative models

Generative models models the distribution of individual class Naive Bayes, Bayesian Networks and Hidden Markov models are commonly used generative models.