← Machine Learning Foundations
Free
Supervised Learning Mental Models
Features, labels, hypothesis class, and the train/serve gap.
Cheatsheet — Supervised Learning
| Concept | Remember |
|---|---|
| Goal | Generalize to unseen data |
| Baseline | Always beat a dumb heuristic |
| Split | Train / val / test (or CV) |
| Leakage | Instantly invalidates metrics |
Start simple: linear/logistic → trees → only then deep models.