ptitprince 0.3.1

A Python implementation of Rainclouds, originally on R, ggplot2. Written on top of seaborn.

PtitPrince

A Python implementation of the "Raincloud plot"! See: https://github.com/RainCloudPlots/RainCloudPlots

Installation

You can install it via pip:

pip install ptitprince

or via conda:

conda install -c conda-forge ptitprince

or directly from GitHub

pip install git+https://github.com/pog87/PtitPrince

Academic use

To cite Raincloud plots please use the following information:

Allen M, Poggiali D, Whitaker K et al. Raincloud plots: a multi-platform tool for robust data visualization [version 2; peer review: 2 approved]. Wellcome Open Res 2021, 4:63 (https://doi.org/10.12688/wellcomeopenres.15191.2)

History of this project

This is a Python version of the "Raincloud plot" (or "PetitPrince plot", depending on the orientation) from R (under ggplot2) to Python. The Raincloud plot is a variant of the violin plot written in R ggplot2 by Micah Allen. I found a tweet asking for a Python version of the Raincloud plot, and I agreed to give it a try. Alas, the Python version for ggplot2 (plotnine) does not allow to create new styles in a comfortable way. So I decided to write this package using the seaborn library as a foundation.

Then I replicated the plots from the original post by Micah Allen, in Jupyter Notebooks and transformed that code into a Python package.

Since then, the package has received some publicity, and is for example listed in "awesome-python-data-science".

Changelog

See CHANGELOG.md for detailed version history.

v.0.3.1

* Modern packaging with pyproject.toml
* Comprehensive test suite with 89% coverage
* Python 3.9-3.12 support
* Removed default palette to avoid seaborn 0.14 warnings
* Pre-commit hooks and code quality improvements

v.0.3.0

* Seaborn 0.13.2 compatibility
* Fixed raincloud component alignment with hue
* Improved dodge alignment and axis labels

v.0.2.x

* PtitPrince now relies on seaborn 0.10 and numpy >= 1.13
* kwargs can be passed to the [cloud (default), boxplot, rain/stripplot, pointplot]
                 by preponing [cloud_, box_, rain_, point_] to the argument name.
* End of support for python2, now the support covers python>=3.6

Plans for the future:

• ask seaborn mantainers to add this new plot type (not gonna happen)
• add a "move" option in seabon to control the positioning of each plot, as in ggplot2. (either, added in ptitprince)
• get RainCloud published (done!)
• add logarithmic density estimate (LDE) to the options for the cloud
• add the repeated measure feature