Dimensionality Reduction

What is Dimensionality Reduction?

Dimensionality reduction projects high-dimensional data into fewer features while preserving structure needed for visualization, compression, or modeling—examples include PCA, t-SNE, UMAP, and autoencoders.

It can remove noise and speed training, but aggressive reduction may discard rare signals that matter for safety-critical classes.

Authoritative reference: scikit-learn decomposition