statsjunk

A collection of statistical odds and ends.

Framework-free Python functions for common statistical tasks — correlation, regression, spatial autocorrelation, electoral fragmentation indices, and prediction-model sample size calculations. Every function takes plain arrays and returns a typed Pydantic result model; invalid input raises ValueError rather than failing silently or returning NaN.

Where to go next

• Installation — one pip install, no optional extras
• Quick Start — run a correlation and a sample-size calculation in a few lines
• API Reference — every public function and class, with the paper it implements

Modules

One folder per statistical domain, one module per technique inside it. Everything is also re-exported from the top-level statsjunk package, so from statsjunk import compute_pearson_correlation always works regardless of where a function actually lives.

Module	Description
correlation	Pearson and Spearman correlation with confidence intervals, plus the shared Fisher z-transformation
regression	Simple and multiple linear regression, with standardized coefficients and VIFs
spatial	Global Moran's I spatial autocorrelation
elections	Laakso-Taagepera effective number of parties/candidates
samplesize	Minimum sample size for continuous, binary, and survival prediction models (a Python port of R's pmsampsize)