Not every matrix has nice eigenvectors — some aren't even square. The singular value decomposition (SVD) is the grand generalization that works for every matrix, with no exceptions. It says any linear map, however complicated, is really just three simple moves in a row:

First a rotation (V^{\mathsf{T}}), then a pure axis-aligned stretch by the singular values in \Sigma, then another rotation (U). Every matrix is a rotate–stretch–rotate sandwich.

A circle becomes an ellipse

Here's the picture that says it all. Any matrix turns the unit circle into an ellipse. The lengths of the ellipse's two semi-axes are the singular values, and those axes are always perpendicular. Apply the transformation below and watch the circle swell into its ellipse, principal axes marked.

Why the SVD rules data science

The singular values rank the directions by how much the matrix stretches them — so the largest few capture most of what the matrix does, and the tiny ones can often be thrown away. That is exactly how you compress an image, denoise data, build recommender systems, and run principal component analysis — keep the strong directions, drop the weak ones. The SVD is often called the most useful single idea in applied linear algebra, and it's the perfect capstone for everything you've built here: rotations, stretches, eigenvectors and orthogonality, all in one decomposition.