Skip to content

Lab 32 — Final Submission & Exam Review

Session: Week 16 · Session 32 · 90 min CLO: CLO-1, CLO-2, CLO-3 (all) Difficulty: Advanced

Learning objectives

By the end of this lab you can:

  1. Audit your project repo against the final submission checklist.
  2. Self-assess your work against the project rubric honestly.
  3. Review all three modules for the final exam in one pass.
  4. Hand in a submission that a stranger can run.

Problem statement

Today is the last session: final project submission AND final exam. This lab is the pre-flight checklist + exam review. Do the checklist before submitting — each unchecked item is a mark you are choosing to lose.

Dataset requirements

Your own project (final state). No new data.

Step-by-step tasks (use the checklist first — 30 min)

  1. Repo audit — mark ✓/✗ for each:
  2. README.md names team, question, dataset, license
  3. requirements.txt present with version header
  4. notebooks run top-to-bottom (Restart & Run All)
  5. every randomized call seeded (SEED or random_state)
  6. results reproducible: the metric in the README matches the notebook's last run
  7. AI disclosure included (from Lab 30)
  8. data sheet / provenance included
  9. presentation.md from Lab 31 committed
  10. Fix anything ✗ — at minimum: re-run notebooks, update README numbers, add the disclosure if missing. Commit and push.
  11. Self-assessment (15 min): score your project 1–5 on: question clarity, data transparency, method fit, results communication, limits. Write one sentence per score, and one sentence on what you would change with one more week.
  12. Final commit: git log --oneline should tell the project's story; make one last commit ("final submission") if anything is uncommitted.

Exam review — Module A (10 min)

In one markdown cell, write from memory (then check): - the 4 pandas selection tools and when each is used, - the difference between isna() and dropna(), - what groupby(...).agg(...) returns and its shape.

Exam review — Module B (10 min)

Write from memory: - the 4-step sklearn API, - what train_test_split protects against and what random_state does, - the meaning of accuracy vs. precision/recall, MAE vs. R².

Exam review — Module C (10 min)

Write from memory: - the agent loop (reason → call → observe → decide), - two guarantees the registered-tools boundary provides, - the AI-use disclosure policy's three requirements (from assessment-plan.md), - the n8n three-part structure (trigger → processing → output).

Starter code

## Submission checklist
| Item | ✓/✗ |
|---|---|
| README: team, question, dataset, license | |
| requirements.txt with version header | |
| notebooks run top-to-bottom | |
| all random calls seeded | |
| README metric matches notebook | |
| AI disclosure included | |
| data sheet included | |
| presentation.md committed | |

## Self-assessment (1-5)
- Question clarity: ...
- Data transparency: ...
- Method fit: ...
- Results communication: ...
- Limits: ...
- With one more week I would: ...

Expected output

  • Checklist with all 8 items ✓ (or ✗ with the fix noted and done).
  • Self-assessment with scores + the "one more week" sentence.
  • Final commit pushed; git status clean.
  • Three exam-review cells written from memory first, then corrected.

Questions

  1. Which checklist item protects you from the "works on my machine" grade complaint? Why?
  2. Why must the README's headline metric match the notebook's last run?
  3. During an exam, what would you check first when a model errors with a shape mismatch in train_test_split?
  4. You used AI to write your EDA code. What does the disclosure require, and what is the consequence of omitting it?
  5. Name the single most valuable habit from this course for your next data project — and why.

Challenge task

Write the course retrospective (5 sentences): one thing that clicked early, one thing you'd warn a next-semester student about, one skill you'll use immediately in another course, and one skill you want to go deeper on. Commit it as retrospective.md in your repo — your future self will thank you.

Marking rubric

Criteria Max Notes
Checklist completed (8 items) 8 ✓/✗ per item
Fixes executed 4 ✗ items addressed
Self-assessment with scores 4 5 scores + sentence
Final commit + clean status 3 log tells the story
Module A review 4 correct, from memory
Module B review 4 correct, from memory
Module C review 4 correct, from memory
Answers to questions 4 Q1, Q3, Q4 correct
Challenge: retrospective 3 5 sentences, committed
Total 38