Linear regression

Fit a straight line through data to predict a number, like price from size.

The model is a line: prediction = w·x + b. `w` (slope/weight) is how much output changes per unit of input; `b` (bias/intercept) sets the starting height. Training finds the best w and b.