What is ALS (Alternating Least Squares)?

Updated May 16, 2026

Short answer

ALS is a matrix factorization technique that alternates between optimizing user and item matrices.

Deep explanation

ALS fixes one matrix (user or item) and solves least squares for the other iteratively. It is scalable and works well for distributed systems like Spark.

Real-world example

Large-scale movie recommendation systems in Spark.

Common mistakes

  • Ignoring regularization leading to overfitting.

Follow-up questions

  • Why ALS is scalable?
  • What is rank?

More Recommendation Systems interview questions

View all →