Debugging ML Models

1. Understand business context and objective for the ML model
2. Check featurization and modelling are done correctly with no leakage
   • Inspect that suitable model architecture is used
3. Inspect data and label distributions
   • Are there drifts?
   • How does it perform against random? (Check F1 score)
4. Inspect data and label quality
5. Inspect features used
6. Inspect poor hyperparameter choice