Matplotlib 2 0 or "One does not simply change all the defaults"

on YouTube ↗35m

" summary

Matplotlib 2.0 actually changes defaults and documents what changed: a new Vega 10 color cycle with C0–C9 color notation, new perceptually linear colormaps (viridis magma inferno plasma) replacing jet, scatter and legends following the prop_cycle, dash patterns scaled with linewidth, a categorical-plotting hook for lists of strings, a image pipeline rewrite that fixes a 15‑year color resampling bug, better pandas/date handling, DejaVu default font, interactive/IPy widget backend work and many 1.5.x/2.0 beta fixes.

AI-written summary · about

" chapters (17)

1. 0:00Matplotlib 2.0 delay
2. 3:34Defaults rabbit holes
3. 4:59New defaults and styles
4. 7:50Color cycle C0-C9
5. 9:56Cycler utility
6. 11:46Property cycle scatter
7. 14:38Pandas integration
8. 16:44Default colormap issues
9. 18:54Image interpolation bugs
10. 20:58Default font DejaVu
11. 22:46XKCD colors additions
12. 23:49IPython auto-refresh
13. 25:34OSX backend overhaul
14. 27:25Matplotlib governance
15. 29:50Community contributions
16. 31:32Domain-specific plotting
17. 32:39Q&A

" related talks

• 29mInteractive visualization for the curiousEdward Schofield · PyCon Australia 2015
• 37mData visualization with plotly py: Version 3 and beyondJon Mease · PyData DC 2018
• 30mFrom plotting to chart porn - Creating slick charts with MatplotlibTamir Lousky · PyCon Israel 2017
• 42mRefactoring PyChartJacob Hallén · EuroPython 2011
• 42mInteractive Data Analysis: Visualization and BeyondJeffrey Heer · PyData Seattle 2017
• 38mBuilding a maintainable plotting library-PyData NYCColin Carroll, Hannah Aizenman, Thomas Caswell · PyData New York City 2019
• 22mVEGAS: The missing Matplotlib for SparkRoger Menezes · Scale By the Bay 2017
• 42mAltair Declarative, Statistical Visualization for PythonBrian E. Granger · PyData San Francisco 2016