Difficulty Distribution
Skills You'll Practice (9)
All 3 Scenarios
Click any scenario to practice it. No account required.
Model Predictions Getting Worse
Your customer churn model has been in production for 6 months. Initial accuracy was 87%. This month it dropped to 71%. Your model retrains weekly. Marketing changed their campaign strategy 2 months ago. What do you investigate first, and what's your action plan?
Design Recommendation System
You're building a recommendation system for a music streaming service (100M users, 50M songs). Users want personalized daily playlists. Walk me through: what algorithm(s) you'd use, how you'd handle cold-start, how you'd measure success, and infrastructure requirements.
Model Shows Bias Against Minority Group
Your loan approval model is 89% accurate overall. But it approves loans for Group A at 78% and Group B at only 41%. Both groups have similar creditworthiness in your data. The company launches in 6 weeks. What do you do?