32  Decomposition and Dimensionality Reduction

32.1 Learning objectives

• Explain why a lower-dimensional representation may be useful.
• Interpret PCA scores, loadings, and explained variance.
• Distinguish a visualization of the data from evidence about its structure.

32.2 Why reduce dimensionality?

32.3 Centering, scaling, and feature selection

32.4 Principal component analysis

32.5 t-SNE

32.6 UMAP

32.7 Interpreting projections

32.8 Demo: Decomposition with rtemis

32.9 Hands-on practice

32.10 Checkpoint

32.11 Further study