ComStats

Matrix broadcasting combinatorial statistics. The reasons for having a separate repository for this stems from the communication here.

Installation

pip install ComStats

Example

This library enables numpy matrixes to be passed into statistical functions for optimised recursive comparisons across dimensions.

>>> import numpy as np
>>> from ComStats import comstats as cs
>>> input_set = np.array([
        [1, 0, 0, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1],
        [2, 1, 2, 3, 3, 0, 1, 0, 0, 0, 4, 1, 2, 4, 4],
        [1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0],
        [3, 0, 1, 3, 0, 0, 2, 1, 2, 3, 3, 1, 0, 0, 2]
    ])
>>> p_values, scores = cs.t_test(input_set)
>>> p_values
array([[1.        , 0.00353093, 1.        , 0.00961514],
       [0.00353093, 1.        , 0.00353093, 0.43711409],
       [1.        , 0.00353093, 1.        , 0.00961514],
       [0.00961514, 0.43711409, 0.00961514, 1.        ]])
>>> scores
array([[ 0.        , -3.38132124,  0.        , -2.88675135],
       [ 3.38132124,  0.        ,  3.38132124,  0.78881064],
       [ 0.        , -3.38132124,  0.        , -2.88675135],
       [ 2.88675135, -0.78881064,  2.88675135,  0.        ]])

Assuming the input_array if of the form: [A, B, C, D] then the resulting matrix is of the form: [[AA, AB, AC, AD], [BA, BB, BC, BD], [CA, CB...], ...].

To convert in and out of pandas DataFrames, see here.

Available functions:

• t_test() # unweighted/weighted paired/unpaired
• percentage_t_test()

See test/test_stats.py for usage and parameter variation.