Teaching_Guide

Teaching guide

Everything needed to walk into a room and teach this course without re-reading a 120-slide deck first.

File	What it is
semester_plan.md	The eleven sessions on one page: what to teach when, what the assessment rhythm is, what to do if you fall behind
runsheets/	One page per session (lecture_NN.md), plus one per self-study module (module_aN_*.md): timing blocks, what to cut first, which exercises to run live, the misconceptions to pre-empt. Not published — git-ignored, because they map exercises onto exam problems; instructors request them from the author
slide_index.md	Every deck's sections with page ranges, every exercise with its page, generated from the compiled PDFs; each deck's appendix is listed but marked optional rather than given a time budget
before_class.md	The ten-minute checklist for the morning of a lecture
handouts/	Printable two-up versions of every deck. Not published — build output, git-ignored; run make handouts to generate the folder
make_index.py, check_decks.py, check_runsheets.py, check_notebooks.py	The generators behind the index, the deck health check, the runsheet page-reference validator, and the notebook re-execution check (make notebooks)

The one command you need

From the repository root:

make            # regenerate figures, rebuild any deck whose source changed, refresh the index
make check      # page counts, slides that overrun the frame, and stale runsheet page references
make runsheets  # every runsheet page reference with the slide title it lands on
make handouts   # printable 2-up PDFs for every deck
make advanced   # the nine self-study module decks (kept out of `make all`)
make notebooks  # re-execute all 25 notebooks and diff against their stored outputs
make docs       # build the documentation site locally

make help lists the rest. Everything is incremental — a deck is only recompiled if its .tex is newer than its .pdf, and the two precourse decks also rebuild when their make_figures.py changes.

How the pieces fit

Chapters/chapter_NN/chapter_NN.tex   the deck you project
        │
        ├─ make ──────────► chapter_NN.pdf        (two pdflatex passes)
        ├─ make handouts ─► Teaching_Guide/handouts/chapter_NN_handout.pdf
        └─ make index ────► Teaching_Guide/slide_index.md   (page numbers)

Chapters/chapter_NN/chapter_NN_lab.ipynb    what students run
        └─ make notebooks ► re-executed and diffed against its stored outputs

Chapters/Advanced/advanced_NN_topic/        nine self-study modules (make advanced)
Mock_Exams/                                 assessment (kept out of git)

A note on the appendices

Every deck ends with an appendix holding its optional, more advanced material — formal derivations, the heaviest worked exercises, side topics. It is deliberately outside the timed plan: the runsheets stop where the appendix begins, and slide_index.md marks it optional. Three ways to use it:

• Assign it. Every exercise in an appendix has its full solution behind it, so students can work it alone.
• Reach for it in the room when a cohort wants the derivation — each runsheet has an "In the appendix (optional)" section naming what is there and which board moments moved with it.
• Ignore it. No slide in the main flow depends on an appendix slide.

A note on the two precourse sessions

chapter_00 (statistics refresher) and chapter_00b (notation, logs and odds, likelihood, the Python patterns) are optional and sit before Chapter 1. They exist because the ten chapter decks assume all of it silently. Two ways to use them:

• Taught, as one or two extra sessions in the week before the course starts — worth it for a cohort with mixed backgrounds.
• Set as self-study, pointing students at the twelve-question self-check on slide 8 of chapter_00. Students who answer nine or more can skip it.

Either way, tell students the material is assumed from Chapter 1 onwards. The commonest cause of a struggling student in this course is not the machine learning — it is a rusty grasp of standard errors.