Sixteen weeks, one river, one standard: an analysis anyone can check.
One dataset, sixteen weeks, and the habits that outlast the syntax.
Plot the river's record; count its events; first thresholds defended.
Tested functions; grids; the rain-to-peak lag measured.
Figures with claims; regressions with residuals; tools you built yourself.
Verified AI edits; validated maps; a model diagnosed honestly.
The syntax was never the point: each week's question needed the previous week's skill.
?
In Week 6 you measured the rain-to-peak lag. Which THREE earlier skills did that measurement quietly require?
Tiny tests, magnitude anchors, two routes to every important number.
Residuals as diagnosis; unseen data as the judge; calibrated verbs.
Sources cited, assistance logged, work explainable line by line.
Libraries churn; these habits are the skill.
The checklist, item by item, marked right now.
| Check | How |
|---|---|
| Runs for a stranger | Fresh clone, Restart and Run All succeeds |
| Figures complete | Labels, units, legend, caption on every one |
| Data cited | Source, URL, access date in the README |
| Statements match reality | Contribution statements name real work |
| AI log complete | Or explicitly empty |
| One number traced | A conclusion value followed back to raw data by hand |
Ten minutes today saves a grade on Friday.
?
Which checklist item is your group most likely to fail today, and what is the fix before Wednesday?
How the last week runs, and what the course was actually about.
| Component | What strong looks like |
|---|---|
| Question and motivation | Specific, answerable, honestly framed |
| Method | One method, applied correctly, validated |
| Figures | Each makes one claim visible |
| Interpretation | Limitations named without prompting |
| Disclosure and verification | Log complete; checks described |
| Individual understanding | Any member explains any section |
Computation is how modern Earth science asks whether its own claims are true.
Sixteen weeks of checks: asserts, anchors, residuals, holdouts, logs.
A reproducible analysis is an argument anyone can check. Hold your own work to it first.
12 labs, 26 notebooks, one final project, every one reproducible.
7 years of weather, 11 years of river flow, 17 stations, a century of water quality.
Functions with tests, maps with validation, a model with an honest diagnosis.
| When | What |
|---|---|
| Wednesday, Dec 9 | Presentations, part 1: order posted on the course site |
| Friday, Dec 11 | Presentations, part 2 |
| Friday, Dec 11 | Final package due: notebook, figures, README, statements, AI log |
Submission path and presentation order are posted on the course site.
?
A committee member asks: how do you KNOW your main number is right? Which two habits from this course answer them? Practice the answer out loud this week.