The bumps in the road: A retrospective on my data visualisation mistakes

on YouTube ↗28m

" summary

Describes moving from messy Jupyter notebooks to a reproducible Python pipeline using matplotlib, Conda and GitHub Actions that auto-runs every function prefixed plot_ and assembles a single Gallery.pdf artifact. Shows concrete fixes: unit tests that catch extra Y columns, CI failures that alert broken figures, and stress-testing by multiplying input size 100x to find memory limits. Demonstrates separate src/data layout, automated nightly reports, and that the gallery generation reliably detects regressions.

AI-written summary · about

" chapters (13)

1. 0:00Introduction and setup
2. 1:09Why data visualization is hard
3. 4:00Know your audience first
4. 6:06Setting up the project structure
5. 7:29Design principles for clarity
6. 10:17Customize default plots
7. 12:03Testing data visualization code
8. 13:07Project structure deep dive
9. 15:56Avoid misleading visuals
10. 18:07Continuous integration for plots
11. 20:55Know your visualization limits
12. 21:59Key takeaways and automation
13. 23:47Q&A session

" related talks

• 01h 27mLeveling Up Your Storytelling and Visualization SkillsGerrit Gruben · PyData Berlin 2018
• 01h 38mEffective Data VisualizationHusni Almoubayyed · PyCon US 2020
• 02h 31mTech Tutorial: Data visualization in PythonJosip Saban · International Data Science Conference 2021
• 01h 28mPython Visualization for Exploration of DataStephen F Elston, Ronald Lopez · PyData San Francisco 2016
• 38mBuilding a maintainable plotting library-PyData NYCColin Carroll, Hannah Aizenman, Thomas Caswell · PyData New York City 2019
• 02h 23mEffective Data VisualizationHusni Almoubayyed · PyCon US 2021
• 35mBad data - anecdotes and examples from the real worldAvishek Panigrahi · PyData NYC 2023
• 43mBehind the Scenes at Gapminder: A Story about Data and VisualizationsJasper Heeffer · Øredev 2018