oddfisher fisher exact test includes the odds ratio parameter, which is equivalent to or parameter from fisher.test in R, where or is hypothesized odds ratio. This parameter is currently not available in fisher exact test in scipy.stats package, and I hope to benefit the python users, looking for python-based fisher exact test with with hypothesized odds ratio parameter.
Please note that this code is based on R fisher-test implementation, relying on the scipy.stats package to achieve equivalency around the negative hypergeometric distribution functions.
- Clone the repo
> git clone git@github.com:happysadderman/oddfisher.git > cd oddfisher > poetry install .
- To run, check out the usage with help menu
> oddfisher fisherexact -h
Altermatively, if you want to use the fisher exact test as stand-alone script, you can run the code using
> python fisher.py fisherexact 1 2 3 4 --odd-ratio 10 > python fisher.py fisherexact 1 2 3 4 # default fisher-exact with no hypothesized odd-ratio
Example code shows how parameter --odd-ratio can be used during the fisher exact test. When --odd-ratio parameter is not used, it defaults to odd-ratio of 1.
> oddfisher fisherexact --odd-ratio 10 1 2 3 4
Results
-------------------
p-value: {'two-sided': np.float64(0.07542579075425782), 'less': 0.07542579075425782, 'greater': 0.997566909975669}
confidence interval at 0.95: (0.008503581019485222, 20.296323344994953)
odds ratio: 0.6937896639529924
Above is equivalent to R fisher.test, with or=10
> d <- matrix(c(1, 2, 3, 4), nrow=2)
> fisher.test(d, or=10)
Fisher's Exact Test for Count Data
data: d
p-value = 0.07543
alternative hypothesis: true odds ratio is not equal to 10
95 percent confidence interval:
0.008512238 20.296715040
sample estimates:
odds ratio
0.693793