A specific quantity, a specific dataset, a specific comparison, three working weeks. That is the whole recipe.
| Item | Due |
|---|---|
| Group proposal (template notebook) | Wednesday workshop, this week |
| Analysis notebook, runs top to bottom | Friday, December 11 |
| 2-3 quality figures + short README | Friday, December 11 |
| Presentations (10 min per group) | December 9 and 11 |
| Individual contribution statements | Friday, December 11 |
| AI usage log, if tools were used | Friday, December 11 |
Fall Break sits in the middle: a solid proposal this Wednesday buys a calm November.
Workable against career-sized, and three tests for any candidate.
How did the 2024 flood's rain-to-peak lag compare with the five largest events of 2015-2025? Specific quantity, dataset in hand, one method.
How does climate change affect Iowa rivers? No dataset settles it; no method is THE method; December arrives first.
In the repo, or one approved download away. This week, not week 15.
A single course method answers it, done well.
A question that cannot fail is a description, not a question.
?
"We will analyze water quality in Iowa." Run the three tests on it. What is the smallest fix that makes it workable?
What you already have, and which lab each method came from.
Iowa City daily and hourly, 2019-2025.
Iowa City daily 2015-2025, plus 17 watershed stations with drainage areas.
Historical water quality (1901-2012), synthetic spatial and grid sets.
Outside data are welcome with instructor approval and a documented download step.
| Method | You built it in |
|---|---|
| Statistics and regression | Group Lab 2 |
| Interpolation and algorithms | Group Lab 3 |
| Spatial prediction (IDW, variogram) | Group Lab 5 |
| Environmental modeling | Group Lab 6 |
| Time-series analysis | Lab 6 |
| Visualization-led analysis | Group Lab 1 |
One method done well beats three sampled.
Roles, ownership, AI disclosure, and the reproducibility bar.
fresh clone -> Restart and Run All -> your figures appear
grader machine, not yours
no hand edits, no missing files
data documented or downloadable
README says what, how, who
One good project sketched, how it is scored, and three milestones.
Do smaller watersheds respond faster? Rain-to-peak lag vs drainage area across 17 stations.
Event detection (Lab 2) + time-series lags (Lab 6) + one regression (GL2).
Lag vs area scatter with fit; two contrasting event hydrographs; station map.
| Component | Weight |
|---|---|
| Research question and motivation | 15% |
| Data cleaning and reproducibility | 15% |
| Computational method | 20% |
| Figures and communication | 15% |
| Interpretation and limitations | 15% |
| AI disclosure + verification of code | 15% |
| Individual contribution and understanding | 10% |
Half the grade lives outside the code: framing, figures, honesty, ownership.
?
Your group's question needs a dataset you have not found yet. What is the decision rule?
Proposal locked, data loaded, one exploratory figure.
Main analysis complete, figures drafted, risks retired.
Polish, README, rehearse, present, submit.
Fall Break is next week: whatever is not locked Wednesday drifts two weeks.
| Session | Purpose |
|---|---|
| Wednesday | Proposal workshop with scope checks |
| Friday | Project studio: data loaded, first figure |
| Week 15 Monday | Advanced-application mini-modules, by class vote |
?
Write your group's question in one sentence, right now. Trade with another group and run the three tests on each other's.