01·Linear regression — intuition

Introduction

Linear regression is the foundation of classical ML and the entry point to supervised learning. In this lesson we build geometric intuition (best-fit line through a cloud of points), algebraic (model y = w·x + b), and statistical (Gauss-Markov assumptions, residuals, BLUE). We show the difference between simple regression (1 feature) and multiple regression (n features as a hyperplane in R^(n+1)), and why "linearity" refers to parameters, not features (polynomial regression is also linear in parameters).