Introduction
Not every relationship is a straight line:
- a car's fuel use rises sharply at high speeds;
- crop yield rises with fertiliser up to a point, then falls;
- the height of a thrown ball follows an arch.
Surprisingly, you can still use linear regression to fit these curves. The trick: create new features that are powers of the originals (x², x³…), then fit an ordinary linear model on them. This is called polynomial regression.