What is matrix factorization in recommendation systems?
Updated May 16, 2026
Short answer
Matrix factorization decomposes user-item matrix into latent user and item factors.
Deep explanation
It learns low-dimensional embeddings for users and items such that their dot product approximates observed ratings. This helps generalize from sparse data.
Real-world example
Netflix Prize-winning recommendation approach.
Common mistakes
- Not handling missing values properly.
Follow-up questions
- What are latent factors?
- Why is it powerful?