ALS model settings

Latent factors k: 6 Regularization λ: 0.10 Max ALS iterations: 25

Selected listener

Observed plays (train)—

Top recommendations

Training stats

ALS iteration0 / 25

Train RMSE—

Held-out RMSE—

Matrix size—

Sparsity—

Each ALS sweep alternately re-solves a real regularized least-squares system for every listener row (holding tracks fixed) and then every track row (holding listeners fixed) via normal equations U·Vᵀ ≈ R. Recommendations for the selected listener are recomputed every sweep from the current factors.