Logistics Regression

Classification model

the probability of a binary outcome by applying a logistic (sigmoid)

Loss Minimization View of ML

Three design decisions

• • Model family: What are the candidate models 𝑓? (E.g., linear functions)
• • Loss function: How to define “approximating”? (E.g., MSE loss)
• • Optimizer: How do we optimize the loss? (E.g., gradient descent)

---

Linear Functions for (Binary) Classification

• Input
• Classification

NLL loss

Classification

Multi-Class Logistic Regression

NLL