wunder beta📱 How Algorithms Predict You
How a wall of numbers learns to guess what you'll watch next.
What you’ll learn
- Understand the sparse rating matrix and the main strategies for predicting preferences.Recommenders fill a sparse user-item grid using collaborative filtering, content-based filtering, and matrix factorization's latent factors.
- Understand cold start, implicit feedback, model ensembles, and the feedback loop's effects.Real systems solve cold start with hybrids, lean on behavioral signals, blend many models, and must guard against filter bubbles.
Questions this course answers
Why is the user-item matrix described as 'sparse'?
No one experiences more than a tiny fraction of all items, so nearly all cells are blank, and the system's job is to predict the missing ones.
What is the core assumption of collaborative filtering?
Collaborative filtering recommends based on the tastes of similar users, betting that past agreement predicts future agreement.
What does matrix factorization learn?
It decomposes the matrix into latent factor vectors for users and items that the algorithm discovers on its own to predict missing ratings.
What is the cold-start problem?
With no interaction history, collaborative filtering cannot place new users or items, so systems fall back on content features or popularity.
Why did Netflix shift toward implicit feedback?
Explicit star ratings were sparse and unreliable; signals like watch time, pauses, and abandonment are more honest predictors of real interest.
What is a key risk of the recommendation feedback loop?
Reinforcing past behavior can trap users in filter bubbles and amplify popular items, so good systems add diversity and exploration.
Grounded in trusted sources
- Matrix factorization (recommender systems), Wikipedia (en.wikipedia.org/wiki/Matrix_factorization_(recommender_systems))
- Shaped, 'Matrix Factorization: The Bedrock of Collaborative Filtering Recommendations' (shaped.ai/blog/matrix-factorization-the-bedrock-of-collaborative-filtering-recommendations)
- Brainforge, 'How Netflix Uses Machine Learning to Create Perfect Recommendations' (brainforge.ai/blog/how-netflix-uses-machine-learning-ml-to-create-perfect-recommendations)
- Frontiers in Computer Science, 'Hybrid attribute-based recommender system with emphasis on cold start problem' (frontiersin.org/journals/computer-science/articles/10.3389/fcomp.2024.1404391/full)
- The Cloud Girl, 'How Netflix Is Building Recommendation Engine with LLM' (thecloudgirl.dev/blog/how-netflix-is-building-recommendation-engine-with-llm)
Every Wunder lesson is built from real, reputable sources — never invented.
Related Science courses
Wunder is a personalized learn-anything platform — tell it any topic and it builds a beautiful, fact-checked course in minutes, with narration, a knowledge check, and a college-style University track.
Browse more Science courses · All topics · Home
© 2026 Wunder Learning LLC · Terms & Privacy