Matrix Factorization (sklearn)
Sparse SVD-based matrix factorization using scipy for collaborative filtering
Matrix Factorization (sklearn) uses scipy's sparse SVD to factorize the user-item interaction matrix. It is a lightweight collaborative filtering approach suitable for implicit feedback data (clicks, views, purchases).
When to use:
- Implicit feedback data (no explicit ratings, just interaction counts)
- When a simple, fast matrix factorization without gradient descent is preferred
- Smaller recommendation datasets
Input: User-item interaction data with interaction counts or binary signals Output: Ranked list of recommended items per user
Model Settings (set during training, used at inference)
N Components (default: 50) Number of latent dimensions for user and item representations.
N Iterations (default: 5) Iterations for the randomized SVD solver.
Inference Settings
No dedicated inference-time settings.