BDA course Aalto: the TeX and Quarto sources behind CS-E5710

A course repository, not a package

The subject here is a course, not a library. BDA_course_Aalto holds the teaching material for Bayesian Data Analysis at Aalto, course code CS-E5710, and the published landing page is at avehtari.github.io/BDA_course_Aalto. Anyone can read it, but the project states its purpose more broadly as well: the same files can be used in other courses.

What makes it worth reading rather than skipping is that the content is not free-standing. The course is built around the third edition of Bayesian Data Analysis by Gelman, Carlin, Stern, Dunson, Vehtari and Rubin, and the repository points at the electronic version of that book as available for non-commercial purposes, at the publisher for hard copies, and at a separate errata page. So the repository is the scaffolding around a book: chapter notes, assignments, a rubric, a demonstration set, and a schedule. Someone wanting to learn Bayesian data analysis from this alone would be reading a syllabus, not a textbook.

The primary language recorded for the project is TeX, which fits the published output of slides and a typeset book chapter, but the working sources you would actually edit are `.Rmd` files under Quarto. That mismatch is worth keeping in mind before you plan an edit: this is a writing repository with an occasional script, not a codebase.

Quarto sources with one file per term and a second institution's variants

The top-level tree explains the layout better than any description could. The build configuration is `_quarto.yml`, the site entry point is `index.Rmd`, and the course body is duplicated per year: `Aalto2023.Rmd`, `Aalto2024.Rmd`, `Aalto2025.Rmd`, `Aalto2026.Rmd`, with `schedule2023.csv` through `schedule2026.csv` alongside. The convention is additive. Old offerings are kept rather than replaced, so a reader has to pick the file that matches their term instead of finding one current file.

Then there is a second institution in the tree. `gsu2023.Rmd`, `assignments_gsu.Rmd`, `project_gsu.Rmd` and `ta_info_gsu.Rmd` name a second delivery of what looks like the same course, with its own assignments, its own project description and its own teaching-assistant information. The project rubric sits separately as `project_rubric.Rmd`, which is the kind of file that gets reused across both.

Supporting directories do real work. `rpackage/` is an R package that the exercises can load, `slides/` holds the slide sources, `figures/` the images, `templates/` the document templates, `extra_reading/` the background list, and `tests/` exists without the tree saying what it checks. `project_work/` holds student-facing work, `FAQ.Rmd` the questions that recur every term, and `BDA3_notes.Rmd` the notes keyed to the book. `demos.Rmd` is the demonstration set. There is also a shell script, `fetch_assignments.sh`, whose purpose the README does not describe.

Where to start reading: index, assignments, rubric, demos, notes

There is nothing to install, and that is the correct first thing to know. No package is published, no build command is given, and the deliverable is a rendered website plus files you can read directly. If you want the polished version, the course web pages are the place to start; if you want the sources, open `index.Rmd` and follow it.

For a student arriving mid-term, the path is short. Find the `.Rmd` file stamped with your year, read the `schedule` CSV with the same year for the week-by-week order, then open `assignments.Rmd` for the exercises and `FAQ.Rmd` for the questions that come up every year. When the assessed project starts, `project.Rmd` describes it and `project_rubric.Rmd` says how it is marked, which is the pairing most self-directed learners never get.

For an instructor borrowing the material, the interesting files are `demos.Rmd`, `BDA3_notes.Rmd` and `extra_reading/`, since those are the parts that adapt to a new cohort. `templates/` is what makes the output reproducible rather than hand-formatted, and the presence of `_quarto.yml` at the top level is the closest thing to a build recipe in the repository. The README does not give a render command, so expect to read that configuration file rather than to follow a documented one-liner.

Updating means cloning and pulling, because there are no releases

The update story is stated more bluntly than most projects state it. The course files change while the course runs, and exercise instructions and slides are updated no later than the Monday of the corresponding week. The recommended way to track that is to clone the repository and pull before checking new files. A ZIP download is offered for people who would rather not learn git, both as a link and as the archive URL for the master branch.

That weekly cadence has consequences. There is no GitHub release to pin, so a course running in week three wants the week three file rather than last term's, and any citation of a specific version of these materials has to be a commit rather than a tag. Two years of schedules sitting side by side is the same fact seen from a different angle: the repository is a running log of offerings, and the history is the versioning.

The default branch is `master`, and the last push was on 2026-09-28 with the repository not archived, so the content is current. The README's build badge points at a travis-ci.org URL for the master branch, while the tree also carries a `.github/` directory, which is a small mismatch worth knowing before you trust either signal as a statement about whether the site still builds.

CC-BY-NC 4.0 for text, BSD-3 for code, the book as a third case

Licensing here is stated in prose rather than in a file, and it is split three ways. Text and videos are under CC-BY-NC 4.0. Code is under BSD-3. The book itself is a separate matter again: the electronic edition is made available for non-commercial purposes, and hard copies come through the publisher.

The non-commercial clause on the text is the part that shapes what a course can do with this. The repository explicitly invites reuse in other courses, and BSD-3 on the code means the `rpackage/` sources can be lifted into something commercial without much friction. Slides, notes and exercises cannot make that same trip without a different answer, and the split between the two categories is not marked file by file in the tree, so anyone reusing the material has to classify each piece themselves. That is a judgement call about a licence rather than an engineering question, and this review does not make it for you.

What is missing is any licence file at the top level. The tree lists `.gitignore`, the README, the course sources and the directories, and no LICENSE entry among them, and the project page shows no licence identifier either. So the terms exist only as a sentence in the README, which is the wrong place to look when a committee asks.

The wrong tool when you want methods rather than a syllabus

Three cases where this is the wrong repository. First, if you want to learn Bayesian data analysis rather than teach or take a course. The repository points at the book and at chapter notes about the book; it does not restate the arguments, and the notes are keyed to a specific edition.

Second, if you need a maintained library. There is no version, no release, and no statement about which toolchain renders the site, and the primary language recorded as TeX is not the language of the source files you would edit. Third, if your course runs on a different calendar. The schedule files are year-stamped per institution, and adapting them means editing the CSV as well as the `.Rmd`, which the second institution's variants show is expected work rather than a rebrand.

One more thing to weigh before adopting. The acknowledgements name around two dozen course assistants, current and previous. That is a sign the material is genuinely maintained by many hands over several years, and also a sign that nobody owns it as software: contributions arrive from people who have finished the course and moved on, and the file that captures their work is prose. A `.DS_Store` committed at the top level is the harmless evidence of how the files actually get edited on macOS. If your requirement is reproducible course infrastructure with a maintenance owner, this repository is not that, and its weekly Monday cadence says so plainly.

Editorial conclusion

Take this repository if you teach Bayesian data analysis and want a course skeleton you can lift, or if you are sitting CS-E5710 and want the same exercises as the class. Do not take it as a methods reference, because the content points at the book rather than replacing it, and do not treat it as a pinned dependency, since the repository has no releases and no licence file at the top level. Clone master and pull on Mondays, and read the CC-BY-NC 4.0 term on the text before reusing slides in a commercial course.

Frequently asked questions

Official sources

1. avehtari/BDA_course_Aalto on GitHub
2. Issues
3. Project website
4. README