Read the online guide. Repair a Python or R plot in VS Code + GitHub Copilot. Use Chat to understand, plan, and run code; inspect plots in Explorer and color-coded diffs in Source Control.
This is Part 2 of the Practical GenAI series, from Lars G. Fritsche and the Fritsche Lab at the University of Michigan.
Practising what we teach: this guide and exercise were developed with substantial AI assistance in writing, coding, testing, and review. They are provided as is, without warranty, to the extent permitted by applicable law. Review the code, agent instructions, permissions, and results before use. Read the AI-assistance disclosure, use guidance, and warranty notice.
Follow the VS Code + Copilot setup to install the tools, sign in, and clone the example repository through Source Control. The download includes code, invented totals, tests, and offline lessons.
Choose Python or R, then follow six lessons through a reviewed local commit and handoff. Bring basic Python/R knowledge; the guide explains Git and requires no typed shell commands on the main path. Plotting itself needs no model API or credentials; Copilot requires account access.
Prefer another interface? Use the Claude Code extension alternative or the complete CLI appendix.
Include accessibility in every figure: contrast, readable labels, and reviewed alternative text.
- Permissions and independent work: approval fatigue, sandbox limits, maintainable code, and avoiding overreliance.
- Agent setup and alternatives: understand Copilot's controls or adapt the exercise to another client.
- Optional R refactoring example: reuse a lab formatting skill or named Codex agent, then verify that results stay the same.
- Repository practices: context, testing, logs, and handoffs.
- Teaching materials: a 45-minute workshop and slides.
- Lab data guidance: institutional requirements before adapting the workflow to study data.
This repository contains the Jekyll website and workshop slides. To preview the site locally with Ruby and Bundler installed:
bundle install
bundle exec jekyll serveOpen the address printed by Jekyll. To check a build:
bundle exec jekyll build --strict_front_matter
python3 scripts/py/check_site.py _siteWith Quarto installed, render the slides with quarto render docs/lab_meeting/slide_deck.qmd and open the resulting HTML. Website and slide output stays out of Git.
The site uses the Fritsche Lab theme and Just the Docs. Synthetic data and reusable templates build on Part 1; see synthetic data provenance.
Use CITATION.cff to cite this guide. GNU General Public License v3.0.