plotting_utils

Elegant wrappers around matplotlib and seaborn, for scientific plotting that stays consistent across figures.

Installation

pip install ac-plotting-ut

Note on the package name. This project was previously published on PyPI as plotting_utils. From v0.2.0 it is distributed as ac-plotting-ut. The import name is unchanged — import plotting_utils as plu keeps working. The old plotting_utils distribution is no longer maintained.

Quick Start

import matplotlib.pyplot as plt
import pandas as pd
import plotting_utils as plu

# Journal-style rcParams
plu.set_rcParams()

df = pd.DataFrame({
    'x': [1, 2, 3, 4],
    'y': [2, 4, 5, 8],
    'group': ['a', 'a', 'b', 'b'],
})

fig, ax = plt.subplots(figsize=(6, 4))
plu.scatter(df, 'x', 'y', by='group', ax=ax)
plu.format_ax(ax, xlabel='X values', ylabel='Y values', reduced_spines=True)

plu.save_best_pdf_quality(fig, figsize=(6, 4), path='.', name='my_plot.pdf')

Every plotting function takes a pd.DataFrame plus column names, draws onto an ax you provide, and returns that ax — so they compose freely with plain matplotlib.

API

🎨 Colors

📊 Plots

🛠 Axis decoration and utilities

Requirements

Python 3.10+. Dependencies (numpy, pandas, scipy, matplotlib, seaborn, scikit-learn, joblib, pillow, statannotations, textalloc) are installed automatically.

Tested on Linux and macOS against Python 3.10, 3.11 and 3.12.

Development

git clone https://github.com/andrecossa5/plotting_utils.git
cd plotting_utils
pip install -e ".[test]"
pytest

Release History

0.2.1

• Fixed bar() silently ignoring categorical_cmap: an unreachable branch meant an explicit palette was never applied and bars rendered in seaborn's default colour
• bar() no longer needs an explicit categorical_cmap=None when by is unset
• Replaced seaborn's deprecated palette-without-hue usage, removed in seaborn 0.14
• Added tests asserting rendered colours, not just that a call succeeds

0.2.0

• Renamed the PyPI distribution to ac-plotting-ut; import name unchanged
• Modern pyproject.toml packaging (hatchling); setup.py removed
• Declared previously-missing dependencies: numpy, pandas, scipy, pillow
• Requires Python 3.10+; verified against numpy 2.5, pandas 3.0, matplotlib 3.11, seaborn 0.13
• Added a test suite (85 tests) and CI across Linux/macOS on Python 3.10–3.12
• Automated PyPI releases via Trusted Publishing
• Fixed volcano() raising ValueError when called with its own default ylim
• Fixed volcano() and rank_plot() label placement failing on non-integer indices
• Fixed rank_plot() passing an invalid positional argument to ax.scatter
• Fixed dotplot and rank_plot being absent from __all__
• violin() no longer needs an explicit categorical_cmap=None when by is unset

0.1.5

• Last release under the plotting_utils distribution name

0.1.4 (2025-09-11)

• Enhanced README with comprehensive documentation
• Added proper package metadata (long_description, author, classifiers)
• Improved PyPI package presentation

0.1.3

• Previous stable release

License

MIT — see LICENSE.