Why should grouped data such as multiple rows per customer use group-aware splitting?
- A Otherwise the same entity appears in both train and validation, causing leakage
- B It balances the folds
- C It speeds up training
- D It is required by cross-validation
Answer
Otherwise the same entity appears in both train and validation, causing leakage
The model memorises the entity rather than learning a generalisable pattern, producing validation scores that do not hold in production.





