Metadata-Version: 2.1
Name: multiple-hypothesis-testing
Version: 0.1.2
Summary: Several methods of combining P-values
Home-page: https://github.com/alonkipnis/higher-criticism-test
Download-URL: https://github.com/alonkipnis/higher-criticism-test
Author: Alon Kipnis
Author-email: alonkipnis@gmail.com
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Description-Content-Type: text/markdown
License-File: LICENSE

# MultiTest -- Global Tests for Multiple Hypothesis

MultiTest includes several techniques for multiple hypothesis testing:
- ``MultiTest.hc`` Higher Criticism
- ``MultiTest.hcstar`` Higher Criticism with limited range 
- ``MultiTest.hc_jin`` Higher Criticism with limited range proposed by Jiashun Jin
- ``MultiTest.berk_jones`` Berk-Jones test (actually -log(bj)) 
- ``MultiTest.fdr`` False-discovery rate with optimized rate parameter
- ``MultiTest.minp`` Minimal P-values (Bonferroni style inference) (actually -log(minp))
- ``MultiTest.fisher`` Fisher's method to combine P-values
 
All tests rejects for large values of the test statistics. 

## Example:
```
import numpy as np
from scipy.stats import norm
from multitest import MultiTest

p = 10
z = np.random.randn(p)
pvals = 2*norm.cdf(-np.abs(z)/2)

mtest = MultiTest(pvals)

f = mtest.fisher()
bj = mtest.berk_jones()

print(f"Fisher = {f[0]}, degrees of freedom = {f[1]}")
print(f"Berk-Jones = {bj}")
```
