To do before Fall 2026 term starts

Running list of changes needed to move this site from the Winter 2026 copy to Fall 2026.

Environment / packages

• Render playground/package-check.qmd on the Maize RStudio server after the update and review the availability table + functional checks. Also render it to PDF (quarto render playground/package-check.qmd --to pdf) to confirm the LaTeX toolchain.
• Known to watch: gghighlight breaks with newer ggplot2 (fails on my laptop’s ggplot2). Confirm the server’s ggplot2 / gghighlight combination works, since it’s used in the plotting slides.

Deployment / repo

• Move repo from aluby/stat220-f26.github.io to a stat220-f26 GitHub org, served at stat220-f26.github.io. Until then, absolute links to https://stat220-f26.github.io/... (data reads, activity/file links) and https://github.com/stat220-f26/... will not resolve.
• Update git remote after the move.

General software setup

• Copy the Gradescope course from last term first. Duplicate the Winter 2026 Gradescope course to carry over the assignment/rubric structure, then update the course name/term and all assignment dates for Fall 2026.
• Then link the new Gradescope course to Moodle — do this after the copy, not before, since the Moodle link attaches to a specific Gradescope course. Confirm the roster syncs correctly once linked.

Content that still points at old terms

• course-syllabus.qmd has freeze: true — the HTML and Typst/PDF syllabus will not rebuild on a normal quarto render. Force it: quarto render course-syllabus.qmd.
• “Spend a Token” Google Form in course-links.qmd:12 (https://forms.gle/aa59m9WVn7u2Bhwo8) — confirm it’s the Fall 2026 form, not last term’s.
• Google Sheet IDs (left as-is for now, update when ready):
  • index.qmd:17 — schedule sheet (currently the W26 sheet; homepage renders last term’s schedule until swapped). Note index.qmd has freeze: false, so it re-reads on every render.
  • portfolio/final.qmd:65 — final project ideas/schedule sheet.
  • portfolio/final.qmd:215 — group assignment spreadsheet.

Older-term leftovers (pre-W26 copies)

• slides/02/slides02.qmd:169 — stat220-s25.github.io
• slides/04/slides04.qmd:107 — stat220-w25.github.io
• slides/slides-cheat-sheet.qmd:152 — github.com/stat220-w25
• activities/10-combining.qmd:10 — github.com/stat220-w25/activities
• activities/25-shiny-design.qmd:97 — sta199-f25.github.io image
• files/23-app1-starter.R:12 — stat220-s25.github.io/data/manager-survey.csv
• lab-quiz/lab-quiz-02.qmd:212 — stat220-s25.github.io/data/lq2_plot_data.csv

w26-named files (rename + refresh content)

• slides/img/w26-schedule.png and duplicate img/w26-schedule.png — last term’s schedule screenshot; used by slides/01/slides01.qmd:309 (../img/w26-schedule.png). Rename to f26-schedule.png, replace the image, update the reference.
• files/app-w26-sol.R
• slides/23/manager-survey-app/app-w26.R
• slides/23/manager-survey-app/app-w26-sol.R
  • Referenced from activities/23-shiny1.qmd, activities/24-shiny2.qmd, slides/23/slides23.qmd. Renaming is optional (they work as-is); internal URLs were already repointed to stat220-f26.
• slides/01/slides01.qmd — bare stat220-f26.github.io / github.com/stat220-f26 text at lines ~240-241, 284, 293, 433 (already updated by the term rename; sanity-check they read correctly).

Teaching-example dates / years (no render impact — update only if desired)

• activities/21-scraping.qmd, slides/20/slides20.qmd, slides/21/slides21.qmd — Carleton catalog term codes (25WI, 26SP) and box-office year/2026 examples.
• Date-parsing examples using "2025-01-29" etc. in slides/11, slides/13, slides/22.
• activities/10-combining.qmd — dataset described as “Fall 2024 and Winter 2025 terms”.

Syllabus / logistics to review

• course-syllabus.qmd — meeting times, room (CMC 306), office hours, office (CMC 223 / CMC 307) for Fall 2026.
• _quarto.yml:27 — Moodle course id set to 50418; confirm.
• Term schedule dates (breaks, no-class days, finals) — these live in the Google schedule sheet, not the repo.

Portfolio projects: 4 → 3 (grading load)

New lineup: P1 (rewritten) → P3 (unchanged) → P4 (unchanged). Cut P2 (“Wrangling weather forecasts”). P3 covers functions/iteration, P4 covers interactivity — both kept as-is.

• P1 rewrite — “Real or fake?” Draft is in _playground/portfolio-1-draft.qmd (data-forensics argument about a Kaggle dataset; replaces the old “pose two research questions” P1). Review, then move to portfolio/portfolio-1.qmd.
• Decide numbering. Either renumber P3→P2 and P4→P3, or keep the labels 1 / 3 / 4. Renumbering touches portfolio/portfolio-*.qmd filenames + titles, the sidebar in _quarto.yml (portfolio section isn’t currently listed there — check), the schedule sheet, and any cross-references in slides/activities.
• Delete portfolio/portfolio-2.qmd and its data (data/weather_forecasts.csv, forecast_cities.csv, outlook_meanings.csv) once the cut is final. Check the schedule sheet and slides for links to it.
• “2 of 4 must be excellent for an A” → “2 of 3.” Appears in the portfolio-2.qmd callout (being deleted) and the new P1 draft; check course-syllabus.qmd and the schedule sheet too. Confirm 3 projects still leaves enough revision / token slack.
• Joins lose portfolio-level assessment with P2 gone. Add “combines data from more than one table or source” as an explicit wrangle criterion on portfolio/final.qmd’s rubric. (HW4–5 already cover joins heavily.)
• Due date: move P1 a little later than the old P1 — aim week ~3–4 — so students have basic dplyr for summaries and cross-tabs. Keep ≥2 weeks between P1 and P3 so you’re not grading two portfolios back-to-back.
• Pairs? Old P1 was individual (good early diagnostic + low stakes). Allowing pairs on the new P1 would further cut submissions — decide.
• Class scaffolding needed. The new P1 assumes a “tells” toolkit that isn’t currently a lecture: distribution smell tests, impossible/contradictory records, duplicates, round-number clustering, decimal precision, correlation structure, categorical balance, temporal regularity, Benford (optional), external benchmarks. Add an EDA / data-quality session near the ggplot weeks.
• Vet the Kaggle shortlist before publishing (see below) — confirm the slugs still resolve, the synthetic ones aren’t self-labeled “synthetic” on the data card, and the real ones still carry the messiness that makes them arguable. Rebalance the mix if needed.
• Pre-existing P1 bug: the current portfolio/portfolio-1.qmd Submission section says “two visualizations (replication and improvement)” — leftover from another project, contradicts the rest of P1. Gone in the rewrite; just noting it’s a known issue.
• P4 revisit menu shrinks to P1(new) / P3. An interactive “data-forensics explorer” over the P1 dataset is a fine P4 base — no change needed, worth mentioning in P4.

Remove / compress homework

• HW7 P4 + P5 — same “lm coefficients per season” problem done as a for loop and as map. Keep one (buys back room on the 9-pt assignment).
• HW6 P2–P4 — three MAC-address string problems on one contrived example. Collapse to one or two; separate_wider_regex() now does most of it in one call.
• HW8 P2 + P3 + P4 (currently the S25 version) — pulls the same ACS data via tidycensus, then again via raw httr2, then wraps the raw version in a function. Cut to tidycensus once + one raw-API problem on a different endpoint.
• HW5 P10 — trim regex gauntlet from 7 parts to ~5. Drop “3+ vowels in a row” and “same first/last letter” (backreference) — puzzle, not practice.
• HW9 P7 (WAI / RMariaDB to scidb.smith.edu) — replace with a DuckDB-backed problem (see Upweight). Depends on Smith keeping a MariaDB server reachable from Carleton; it will fail some semester.

Downweight

• For-loops + manual preallocation (Slides 17) — R4DS 2nd ed (2023) de-emphasizes this; iteration is now across() → map() → “read many files.” Keep one for-loop for the mental model, shift weight to map/across. (Keep more only if Python-loop transfer is an explicit goal.)
• Lexicon sentiment analysis (Slides 15) — only deck with zero homework, and AFINN/bing/nrc is a ~2014 treatment. Either leave it light/class-only, or repoint Slides 14/15 toward TF-IDF (tidytext::bind_tf_idf) + a short “how modern NLP differs” coda.
• ggmap basemaps (Slides 05) — get_stadiamap() now needs a registered Stadia API key (same friction that killed Stamen). HW3 map problems already use map_data() + choropleths, which is fine. Drop the tile-basemap slide or switch to maptiles (no key for OSM/CARTO).

Upweight / add

• DuckDB for the databases unit — swap mdsr::dbConnect_scidb() for dbConnect(duckdb::duckdb()) over local CSV/Parquet. Same dbplyr translation, zero infrastructure, no dependency on someone else’s server. duckdb is already a course dependency. Affects Slides 25–26, Activity 26, HW9 P6–P7. (SQL Murder Mystery is fine as-is.)
• Finish the Quarto-dashboard migration — Slides 22 still library(flexdashboard)s but also shows format: dashboard. flexdashboard is effectively superseded; standardize on native format: dashboard. Also Activity 22 and HW9 P2.
• AI-assisted coding as a skill, not just a policy — one focused treatment near the functions/debugging weeks: verifying LLM-generated R, spotting confident-but-wrong output, attribution.
• gt tables — currently only in the portfolio. Worth one HW problem (pairs with the wrangling or text weeks).
• Parquet / arrow — ~20 min on “bigger than memory” alongside DuckDB.

Same topic, dated function (low effort)

• separate() → separate_wider_delim() / _position() / _regex() — superseded in tidyr 1.3.0 (Jan 2023). Slides 04 and heavily in Slides 08.
• Joins: by = "id" → join_by(id) — dplyr 1.1.0 (Jan 2023). Also show the new multiple-match / unmatched warnings (students hit them anyway). Slides 10.
• group_by() → .by = for one-off grouped summaries — dplyr 1.1.0. Optional, but it’s what current tidyverse code looks like.

---

Done (Winter 2026 -> Fall 2026 migration, initial pass)

• Replaced stat220-w26 -> stat220-f26 across all source (domain, org/repo, org page) — 33 files.
• _quarto.yml: description Winter 2026 -> Fall 2026; Moodle id=48308 -> id=50418; footer Material 2025 -> Material 2026.
• slides/_metadata.yml, activities/_metadata.yml: institute -> Fall 2026.
• course-syllabus.qmd: subtitle -> Fall 2026.
• course-links.qmd: stale org stat220-s25 -> stat220-f26; stale Moodle id=44937 -> id=50418.
• Renamed stat220-w26.github.io.Rproj -> stat220-f26.github.io.Rproj.