Navigation: <-- Part VI: Principles That Transfer (Reflection) | Main Index | Part VIII: Deep Learning -->

---

Part VII: Unsupervised Learning

Supervised learning requires labeled data. This part introduces the complementary setting: no labels, no correct answers. You will see the two main problem types, grouping and detection, and work through the evaluation challenge that defines this domain. The CRISP-DM inner loop applies exactly as before, but with different tools for measuring progress.

---

Nuggets in This Part

#	Nugget	Prerequisites
1	Unsupervised Learning	Supervised Learning · CRISP-DM
2	k-Means Clustering	Unsupervised Learning · Hyperparameter Optimization
3	Anomaly Detection	Unsupervised Learning
4	Isolation Forests	Anomaly Detection · Random Forests

---

Script v1.7 (2026-07-28) · FGN