# Truncated SVD

> Source: https://sukruyusufkaya.com/en/glossary/truncated-svd
> Updated: 2026-05-13T21:01:01.028Z
> Type: glossary
> Category: makine-ogrenmesi
**TLDR:** A truncated singular value decomposition method used for dimensionality reduction, especially in sparse matrices.

<p>Truncated SVD is highly useful for dimensionality reduction in large and sparse data structures. It is especially effective in text mining, recommendation systems, and high-dimensional sparse feature spaces. Like PCA, it produces a low-dimensional representation, but because it does not require centering, it is more compatible with sparse matrices. This makes it especially valuable in practical data science workflows.</p>