Great Tables

Episode Deep Dive

In this episode, Rich Iannone and Michael Chow (both from Posit) join the show to discuss the Great Tables Python library and how it helps developers create attractive, informative, and reproducible data tables. Here are the key points and takeaways:

1. Origins and Motivation

• Rich’s background: Began by developing a similar library for R (called GT), inspired by frequent table creation during his PhD and a deep dive into tabular design references (such as a 1949 U.S. Census style guide).
• Michael’s involvement: Brought experience from data science in both R and Python and helped adapt the original R concepts into Python.

2. Why Tables Matter

• Underestimated Complexity: Tables can be as expressive as charts, but are often overlooked. Even a simple table can convey hierarchy, structure, and storytelling.
• Design Patterns: Emphasis on elements like spanner labels (grouping columns under a shared heading), titles, subtitles, footnotes, and source notes that elevate plain data frames into more polished, communicative tables.

3. Great Tables Library (Python)

• Conceptual Similarities to R’s GT: Great Tables for Python continues the same philosophy—making table creation reproducible, flexible, and visually compelling.
• Data Frame Integration:
  • Pandas and Polars are directly supported.
  • Polars is highlighted for its convenient “selector” syntax and expressions that make styling and transformations more concise.
• Formatting Options:
  • fmt_* methods handle numbers, dates, and even images or flags.
  • Supports markdown in references and footnotes.
  • Nano plots (tiny inline graphs like bar charts) can be embedded directly into table cells for quick visual context.
• Output Destinations:
  • Renders nicely in Jupyter notebooks.
  • Can produce HTML (including fully inlined CSS for email compatibility).
  • Exports to image or PDF.
  • Has a LaTeX option (helpful for academic writing).

Links and Resources

• Great Tables on GitHub: github.com/posit-dev/great-tables-py
• Polars: www.pola.rs/
• Posit (formerly RStudio): posit.co
• Quarto (used to build the Great Tables documentation site): quarto.org
• Tidy Tuesday (referenced as a data inspiration source): github.com/rfordatascience/tidytuesday
• ReactablePy (interactive tables in Python, also mentioned): github.com/posit-dev/reactable-py

4. Contests and Community

• Table Contests: Organized by Posit to encourage creative uses of GT and Great Tables. Participants submit unique, compelling designs—often blending text, images, spark bars, and more.
• Contributing: The Great Tables maintainers encourage pull requests, issues, and new ideas, emphasizing that user feedback fuels new features.

Overall Takeaway

Tables are a powerful medium for data storytelling—far more flexible than many expect. By leveraging a library like Great Tables, developers and data scientists can automate, style, and enrich their tables, all while maintaining the reproducibility and clarity that Python workflows demand.