Module 13 · Responsible systems
Module 13: implement, explain and transfer
Read concepts, then attempt the starter notebook. In a source checkout open notebooks/practice/module13-practice.ipynb. Use the core environment and kernel; the practice checks need NumPy even where the module script does not. Three unfinished functions report not_attempted, not a pass. If a download link does not open, use the offline access route.
Construct recall_denominator
Return the actual-positive count. Predict the result on recall_denominator(8, 2, 11, 30), implement it and call m13_recall_denominator through the notebook check. Then add a changed case and explain which assumption it tests.
Hint for recall_denominator
Recall conditions on actual positives, not predictions.
Construct safe_rate
Return numerator/denominator, or None for zero denominator. Predict the result on safe_rate(8, 19), implement it and call m13_safe_rate through the notebook check. Then add a changed case and explain which assumption it tests.
Hint for safe_rate
Unavailable is not evidence of parity.
Construct wilson
Return a two-number 95% illustrative Wilson interval using learning.wilson_interval. Predict the result on wilson(8, 10), implement it and call m13_wilson through the notebook check. Then add a changed case and explain which assumption it tests.
Hint for wilson
State the conditional independent-binomial assumption before interpreting the interval.
Work with the complete study's results
Open group_audit.csv from the complete study. For the Amer-Indian-Eskimo group, recompute recall with safe_rate and its interval with wilson, then compare the interval with the female group's. What can and cannot be concluded? Compare with the construction feedback afterwards.
Reason beyond the visible checks
Construct confusion counts for a group with no actual positives, and another group with recall 1.0. For each, give the selection rate and the result of safe_rate for recall, and explain why "recall unavailable" is not evidence that the groups are treated equally. The checks cannot mark this; compare your reasoning with the construction feedback.
Transfer to an IT decision
Compare two groups with equal rates and unequal counts. Explain why this does not establish equitable outcomes or justify a group ranking. Run the complete module study and verify one claim against its actual artifacts; keep the evidence the study workflow lists.
Evidence and feedback
- M13.1: Identify stakeholders, historical label limitations, and inappropriate uses.
- M13.2: Fit a declared training pipeline without protected audit columns.
- M13.3: Report group counts, confusion denominators, selection rates, TPR, and FPR.
- M13.4: Explain metric trade-offs, proxy effects, and uncertainty without causal/legal claims.
- M13.5: Write a governance recommendation, including reasons not to deploy.
Use construction feedback after an attempt and recorded experiment feedback for the full workflow. Follow the shared study workflow for evidence, recovery and optional AI disclosure. Official assessment weights remain with your instructor.
Evidence activity: Audit a group comparison
Use group_audit.csv to recompute one group recall from its TP and FN counts and reconcile its denominator. Compare its interval and sample size with another group, then describe a plausible harm and missing contextual evidence. Do not treat overlapping intervals as proof of fairness or use final labels to choose a mitigation.
Follow the four steps in reproduce a result and justify its use. Submit your artifact path, independent calculation, discrepancy or agreement, and bounded use decision in the evidence worksheet. Use your complete reference run; this activity complements the small function checks.