Chapter 8 Logistic Regression
In the previous section, we discussed how discriminant analysis models the joint distribution of the predictors and the class variable. In this section, we focus on logistic regression, which takes a different approach: it directly models the conditional probability of class membership as a function of the predictors. For binary outcomes, this is achieved through the logistic function, which maps a linear combination of predictors to values between 0 and 1.
Logistic regression has the advantage of making fewer assumptions about the distribution of the predictors, provides coefficients that can be interpreted in terms of odds ratios, and can be extended to multiple classes through multinomial formulations. Because of its simplicity, interpretability, and solid probabilistic foundation, logistic regression is one of the most widely used methods for classification.