Data Leakage
10 practice items · needed by 3 roles
Easy 0Medium 5Hard 5
Practice questions
- Scaling before splittingmedium
- Target encoding without leakagemedium
- Several scans per patientmedium
- A feature that is too goodhard
- SMOTE applied before cross-validationhard
- Suspiciously strong churn featuremedium
- Oversampling before cross-validationhard
- Target encoding without leakagehard
- Great backtest, poor live forecasthard
- Joining features to training labelsmedium