15  Dimensionality Reduction: Clustering and Embeddings (todo)

Note

Proposed chapter scope: Unsupervised approaches to making sense of high-dimensional data. Covers clustering (k-means, hierarchical) and dimensionality reduction (PCA, UMAP), with embeddings as a key use case — both word/document embeddings from text and image embeddings from vision models. This is where the conceptual payoff for the “black box” LLM chapter lands.