medstat 1.0.2

Getting started

This project is available on PyPI you can just:

pip install medstat

Quick Example

Load your data in a dataframe using for instance pd.read_csv() or pd.read_excel().

data = pd.read_csv("my_data.csv")

Test a single hypothesis:

>>> medstat.test_hypothesis(data, 'sex', 'age < 30')

{'contingency_table':
age < 30  False  True  All
sex
Female       26    22   48
Male         24     8   32
All          50    30   80,
'test': 'Fisher',
'p-value': 0.06541995357625573,
'significant': False}

Or test many hypothesis at the same time:

result = medstat.analyse_dataset(data,
                                 [('sex', 'age < 30'),
                                  ('sex', 'test_a'),
                                  ('test_a', 'age > 50'),
                                 ])

It prints the output:

-------------------- Test 1 --------------------
Test independence between sex and age < 30.
Use Chi-squared test.
Result is not significant.
p-value: 0.18407215636751517
Contingency table:
 age < 30  False  True  All
sex
Female       21    18   39
Male         29    12   41
All          50    30   80


-------------------- Test 2 --------------------
Test independence between sex and test_a.
Use Chi-squared test.
Result is not significant.
p-value: 0.9539453144224308
Contingency table:
 test_a  negative  positive  All
sex
Female        25        14   39
Male          25        16   41
All           50        30   80


-------------------- Test 3 --------------------
Test independence between test_a and age > 50.
Use Fisher test.
Result is significant.
p-value: 6.392910983822276e-12
Contingency table:
 age > 50  False  True  All
test_a
negative     46     4   50
positive      5    25   30
All          51    29   80

You can also save it to a text file using the file argument.

result = medstat.analyse_dataset(data,
                                 [('sex', 'age < 30'),
                                  ('sex', 'test_a'),
                                  ('test_a', 'age > 50'),
                                 ],
                                file='report.txt')

• Free software: MIT license

• Documentation: https://medstat.readthedocs.io.