Train and Test

Fitting signal, not noise

An interactive lesson with 5 steps, about 5 minutes: you predict, drag and solve, with instant feedback on every answer. It runs in your browser and needs JavaScript.

What you’ll learn

• Judge models on test data they never saw during training.
• Training error falls with complexity; test error is U-shaped.
• Choosing models by test score leaks information: use a validation set.

Steps

1. Wigglier curves
2. The best degree
3. A perfect score
4. Peeking
5. Enough points to pass through