Module 2 Assessment: Data Audit Check
Assessment ID: ML-M02-QA01 Estimated active time: 25-35 minutes Status: Draft
Part A: Concept checks
Answer in one or two sentences.
- What is the unit of analysis in a dataset?
- Why is the target not a feature?
- Why does prediction time matter?
- What is one example of future-information leakage?
- Why can a clean-looking column still be risky?
Part B: Column roles
Classify these columns from learner_progress_audit_v1.csv.
| Column | Role | Reason |
|---|---|---|
record_id | ||
day3_lessons_opened | ||
completed_module1_by_day10 | ||
final_quiz_score | ||
mentor_message_sent_day4 | ||
access_bandwidth_band |
Roles: metadata, feature candidate, target, exclude future information, exclude post-action information, proxy-risk review.
Part C: Audit decision
Write a short data audit note that includes:
- unit of analysis;
- prediction time;
- target;
- valid feature candidates;
- excluded columns;
- missing-value concern;
- proxy-risk concern; and
- continue, revise, or stop decision.
Rubric
| Level | Evidence |
|---|---|
| Pass | Correctly separates target, feature candidates, metadata, future columns, and proxy-risk columns; explains timing and missingness; gives a bounded decision. |
| Revise | Understands the dataset but misses one major timing, missingness, or proxy-risk issue. |
| Not yet | Treats all columns as usable, uses the target as a feature, or ignores prediction time. |