Fast Fisher's Exact Test
Fisher's Exact Test
Simple, fast implementation of Fisher's exact test. For example, for the following table:
|Having the property||Not having the property|
Perhaps we are interested in whether there is any difference of property in selected vs. non-selected groups, then we can do the Fisher's exact test.
Within this folder :
git clone git://github.com/brentp/fishers_exact_test.git pip install .
From PyPI :
pip install fisher
Or install the development version :
pip install git+git://github.com/brentp/fishers_exact_test.git
fisher.pvalue() accepts 4 values corresponding to the 2-by-2
contingency table, returns an object with attributes for left_tail,
right_tail, and two_tail p-values :
>>> from fisher import pvalue >>> mat = [[12, 5], [29, 2]] >>> p = pvalue(12, 5, 29, 2) >>> p.left_tail, p.right_tail, p.two_tail # doctest: +NORMALIZE_WHITESPACE, +ELLIPSIS (0.04455473783507..., 0.994525206021..., 0.0802685520741...)
A simple benchmark that calls the Fisher's exact test 1000 times (in
calling python fisher... iterations/sec: 3000.62526381 calling rpy fisher... iterations/sec: 289.225902364 calling R directly... iterations/sec: 244.36542276
So the cython fisher is up to 10 times faster than rpy or R version.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.