Unsupervised Learning

Status: 🚧 Coming soon — chapters are being written.

Unsupervised learning works without labels. The model finds structure in raw data — groups, low-dimensional representations, anomalies.

What this section will cover

• Clustering — k-Means, k-Medoids, DBSCAN, HDBSCAN, Hierarchical, Gaussian Mixtures
• Dimensionality reduction — PCA, t-SNE, UMAP, autoencoders
• Anomaly detection — Isolation Forest, One-Class SVM, LOF
• Topic modeling — LDA, NMF
• Evaluation: silhouette, Davies-Bouldin, reconstruction error
• Real-world patterns: customer segmentation, embeddings clustering, outlier flagging

Currently available — related material

• Machine Learning → Unsupervised & Clustering
• Machine Learning → Feature Engineering & PCA

A consolidated unsupervised-learning track is in progress.