UMAP
A faster dimensionality-reduction method for visualization that preserves more global structure than t-SNE.
UMAP (Uniform Manifold Approximation and Projection) is a nonlinear dimensionality-reduction technique, like t-SNE, for visualizing high-dimensional data in 2D/3D — but it is faster, scales better, and tends to preserve more of the global structure (relative cluster positions) while still capturing local neighborhoods.
Worked example: reduce millions of embedding vectors to a 2D map with UMAP to explore how documents or users cluster — faster than t-SNE and with cluster spacing that carries somewhat more meaning. Gotcha: UMAP is still a visualization tool, not a faithful metric space — its axes and exact distances are not interpretable, results depend on hyperparameters (n_neighbors, min_dist), and it can create apparent clusters from noise — so, like t-SNE, treat its plots as hypotheses to check, not conclusions.