Logistic Regression (Recap)
Overview Logistic regression is frequently used in classification and predictive anlaytics1. Put it simply, the model estimates the probability (distribution) of (a) certain event(s). Input and Output Input is nothing but the features, just as in the case of linear regression. Denote the input as $\boldsymbol{x}$. Output can (usually) be a vector. We use $\boldsymbol{p}$ to indicate that it is related with probability distribution. For $K$-class classification, the model need to generate $K$ class probabilities, $\{p_k\}_{k=1}^{K}$....