Responsible Operations

Five questions. Choose one answer for each, then submit to see your score and explanations.

The practice target is 80%. You can retry as often as you like, and every lesson stays accessible.

You can practice as a guest. Sign in before starting an attempt to save your scores.

1. Which explanation matches audit leakage before deployment?A. Compare false-positive and false-negative rates across relevant groups and conditions.B. Monitor input shift, outcome performance once labels arrive, and probability calibration.C. Audit timestamps, data lineage, duplicates, and preprocessing to catch leakage before release.D. Use staged evaluation, versioned features, a rollback plan, and evidence across conditions.

2. Which explanation matches evaluate errors across groups?A. Audit timestamps, data lineage, duplicates, and preprocessing to catch leakage before release.B. Compare false-positive and false-negative rates across relevant groups and conditions.C. Monitor input shift, outcome performance once labels arrive, and probability calibration.D. Use staged evaluation, versioned features, a rollback plan, and evidence across conditions.

3. Which explanation matches monitor shift and calibration?A. Monitor input shift, outcome performance once labels arrive, and probability calibration.B. Audit timestamps, data lineage, duplicates, and preprocessing to catch leakage before release.C. Compare false-positive and false-negative rates across relevant groups and conditions.D. Use staged evaluation, versioned features, a rollback plan, and evidence across conditions.

4. For evaluate errors across groups, which result or action fits this practical case: One district receives many unnecessary delay alerts.A. Replay a historical predeparture prediction with only truly available fields.B. Compare recent risk bins with realized delay rates when enough outcomes are known.C. Run both versions on the same future period before switching all traffic.D. Report group counts and inspect whether route availability explains a difference.

5. For retrain and roll back with evidence, which result or action fits this practical case: A replacement model beats the old one in one week.A. Replay a historical predeparture prediction with only truly available fields.B. Report group counts and inspect whether route availability explains a difference.C. Run both versions on the same future period before switching all traffic.D. Compare recent risk bins with realized delay rates when enough outcomes are known.

Review this module from its first lesson →

Responsible Operations

Your result