Here there is no \(y\). You are given \(\mathbf{x}_1, \dots, \mathbf{x}_N\) and asked what structure they contain: which points group together, which items co-occur, which points do not belong. Because there is no ground truth to score against, the hard part of unsupervised learning is rarely the algorithm — it is deciding whether the structure you found is real.
By the end of this unit you should be able to:
Partitioning points into groups: by centroid, by density, and by a hierarchy you can cut at any height.
3 algorithms • K-Means Clustering, DBSCAN, Hierarchical / Agglomerative Clustering 2.2Which items co-occur more often than chance would predict — and how to mine them without enumerating every subset.
2 algorithms • Apriori Algorithm, FP-Growth Algorithm 2.3Finding the points that do not belong, when rarity itself is the only definition you have.
1 algorithm • Isolation Forest| No. | Algorithm | Character | Topic |
|---|---|---|---|
| 2.1 | K-Means Clustering | Centroid-Based • Partitional | 2.1 Clustering |
| 2.2 | DBSCAN | Density-Based • Arbitrary Shapes | 2.1 Clustering |
| 2.3 | Hierarchical / Agglomerative Clustering | Hierarchical • Dendrogram | 2.1 Clustering |
| 2.4 | Apriori Algorithm | Frequent Itemsets • Association Rules | 2.2 Association Rule Learning |
| 2.5 | FP-Growth Algorithm | Tree-Based Mining • Scalable | 2.2 Association Rule Learning |
| 2.6 | Isolation Forest | Anomaly Detection • Tree Ensemble | 2.3 Anomaly Detection |