plotsalot

plotsalot is an independent Python library for statistical visualizations with structured, extractable results. It provides Polars-first analysis workflows and Matplotlib figures inspired by the public behavior of ggstatsplot.

The project is pre-1.0 alpha software. Its classical, robust, Bayesian, categorical, coefficient, and meta-analysis families have passed their recorded M0–M6 technical and statistical gates, but M7 compatibility closure and 1.0 hardening remain. Do not treat a successful computation or plot as automatic validation of a statistical model or a substantive conclusion.

Install

Plotsalot supports Python 3.11 and newer on Linux and macOS. After the first PyPI release:

python -m pip install plotsalot

For development from a checkout:

make setup

R and Docker are development-only oracle tools. They are not package runtime dependencies.

Quick start

import polars as pl

from plotsalot import gghistostats

data = pl.DataFrame({"value": [1.0, 2.0, 3.0, 4.0, 5.0]})
plot = gghistostats(data, "value", test_value=0.0)

print(plot.result.to_dict())
plot.figure.savefig("histogram.svg")

Each plotting function returns a typed container with a Matplotlib figure, named axes, annotations, and a structured result. Analysis and rendering can also be separated so visual changes do not recompute statistics:

from plotsalot import analyze_gghistostats, render_gghistostats

analysis = analyze_gghistostats(data, "value", test_value=0.0)
plot = render_gghistostats(analysis, title="Observed values")
plot.axes["main"].grid(axis="y", alpha=0.2)

Find the right documentation

Supported workflow families

• one-sample histograms and labeled dot plots;
• scatter plots and correlation matrices;
• independent- and repeated-group comparisons;
• categorical bar and pie analyses;
• coefficient and aggregate meta-analysis plots;
• atomic grouped variants, plot composition, and local themes; and
• explicit classical, fixed-trim robust, and approved Bayesian modes where listed in the compatibility matrix.

The implementation is adapted rather than a drop-in port. It uses explicit column names, typed Python results, owned NumPy boundaries, and Matplotlib rendering. Unsupported modes fail explicitly; grouped operations do not return partial results.

Development

make setup
make check
make audit
make build

The default quality gate runs formatting, linting, strict type checking, the test suite, coverage enforcement, and documentation-link checks. Frozen R-oracle and benchmark evidence are verified by the normal tests; regenerating them is an explicit operation:

make oracle
make benchmark

See the documentation index for project plans, decisions, method specifications, and retained verification evidence.

Project identity and license

Plotsalot is an independent project and is not endorsed by or affiliated with the ggstatsplot maintainers. The upstream project is used as a pinned behavioral reference under the adaptation rules recorded in the repository.

Plotsalot is MIT licensed. See LICENSE and ADR-010 for the accepted distribution and upstream-notice rules. Report security concerns according to SECURITY.md.