tkstatistics 0.0.1

Statistics Application using only the Python Standard Library.

tkstatistics

tkstatistics is a small-data statistics desktop application and headless runner built with the Python standard library. Its defining workflow is pre-registration: a confirmatory analysis must match a committed hypothesis plan before the application reveals its p-value.

The project is alpha software intended for teaching and reproducible analysis on modest datasets. It is not a replacement for NumPy, SciPy, pandas, R, or a validated clinical statistics system.

Features

• Tkinter desktop application with CSV import and SQLite project files.
• Reproducible JSON analysis specifications and persisted run artifacts.
• Committed analysis plans, confirmatory-result gating, and dataset audit reports.
• Descriptives, frequency tables, t-tests, nonparametric tests, Fisher's exact test, ANOVA, correlations, and linear regression.
• Histogram, box plot, scatter plot, and normal Q-Q plot rendering with SVG export.
• No third-party runtime dependencies. Numerical routines are cross-checked against established scientific Python libraries in the test suite.

Installation

Python 3.11 or newer is required.

python -m pip install tkstatistics

Tkinter is included with standard Python installers on Windows and macOS. Some Linux distributions package it separately, for example as python3-tk.

Desktop usage

Launch the application with either command:

tkstatistics
python -m tkstatistics

Create or open a .statproj project, import a CSV dataset, then use the Analyze and Graphs menus. Confirmatory analyses follow a two-step flow: first choose Pre-register Hypothesis..., then Run Confirmatory Test....

Headless usage

Run a JSON specification against an existing project:

tkstatistics --run analysis.json --project study.statproj --format text

Machine-readable output is the default and can also be saved to a file:

tkstatistics --run analysis.json --project study.statproj \
  --format json --output run-artifact.json

Pre-registration and auditing are available without the GUI:

tkstatistics --commit-plan plan.json --project study.statproj
tkstatistics --audit dataset_name --project study.statproj

See tkstatistics --help and the examples/mean_hypothesis_demo walkthrough.

Development

uv sync --all-extras --all-groups
make check-ci
make publish-check

The package is MIT licensed. See the release history.