pySHT: statistical hypothesis testing#
pySHT is a Python library for statistical hypothesis testing. It provides literature-traceable implementations for mean, variance, covariance, joint-parameter, distributional, goodness-of-fit, and structured-domain questions, with a consistent result contract across methods.
The package is an extended and independently audited successor to SHT for R. Start with the installation guide, read the user guide, or go directly to the API reference.
Release candidate 0.5.0rc1. Version 0.1.0 remains the latest stable release. This candidate integrates the validated 0.2–0.5 expansion; methods that did not pass their scientific release gate remain private or absent.
Contents#
Getting started — install pySHT and run a first test.
User guide — choose a procedure, prepare data, and interpret the result.
API reference — inspect every public function and result type.
Validation — review formulas, numerical checks, and independent comparisons.
Project — see the roadmap, migration notes, and contribution guidance.
API categories#
[1] Univariate Mean — API
[2] Multivariate Mean — API
[3] Variance — API
[4] Covariance — API
[5] Mean and Variance — API
[6] Mean and Covariance — API
[7] Equality of Distributions — API
[8] Normality — API
[9] Rectangular Uniformity — API
[10] Special Domains — API
[11] Independence — API
[12] Circular Data — API
Categories [0]–[10] preserve SHT 0.1.9’s numbering and scientific
organization. The candidate API has 72 canonical functions: 51 SHT
compatibility functions covering 52 of 54 statistical routine identities and
21 independently cataloged pySHT-native methods. The
migration crosswalk explains the one shared
implementation, the validation-blocked mean CLX and Fisher identities, and the
two R-only adapters; the native catalog records
the additional research methods and blocked candidates.