exact-tests 0.3.0

Accurate and efficient binomial, Hardy-Weinberg equilibrium, and Fisher's exact tests.

exact_tests

This package provides binomial, Hardy-Weinberg equilibrium, and Fisher's exact test functions. (Only binomial test and 2x2 and 2x3 FET implemented right now; the rest is coming soon.) The functions are accurate and efficient:

• High-precision and interval arithmetic are used to ensure likelihood near-ties are correctly resolved. Log-(mid)p-values are available when you don't want to just round tiny p-values down to zero.

• It's relatively expensive to compute a contingency table's likelihood from scratch, but it's cheap when the likelihood of an adjacent contingency table is known. And it's even cheaper to skip a likelihood calculation when the value must be too small to matter. These functions take advantage of the latter possibilities.

Build instructions

PyPI:

pip install 'pip>=20.3'
pip install exact_tests

GitHub:

pip install 'pip>=20.3'
pip install -e 'git+https://github.com/chrchang/stats.git#egg=exact_tests&subdirectory=Python'

Or install from a cloned copy:

# clone repo
git clone https://github.com/chrchang/stats
# go to python folder
cd stats/Python
# install the package
pip install -e .

You can test the package with pytest.

Example usage:

import exact_tests

exact_tests.binom(3, n=15, p=0.1, alternative="greater")
exact_tests.binom(2, 29, 0.1)
exact_tests.binom(2, 29, "0.1")
exact_tests.binom(2, 29, "0.1", midp=True)
exact_tests.binom(100, 10000, "0.000001", logp=True)
exact_tests.fisher([[4, 0], [0, 4]], alternative="greater", midp=True)
exact_tests.fisher([[10000, 20000], [30000, 40000], [50000, 60000]], logp=True)