precise 0.0.2

Online covariance

precise

Online covariance, correlation and precision matrix computations

Install

pip install precise 

Example

from precise.covariance.util import create_correlated_dataset
from precise.covariance.onlineempirical import online_empirical_cov
import numpy as np
from pprint import pprint

data = create_correlated_dataset(10000, (2.2, 4.4, 1.5),
                                 np.array([[0.2, 0.5, 0.7], [0.3, 0.2, 0.2], [0.5, 0.3, 0.1]]), (1, 5, 3))
s = online_empirical_cov(n_dim=data.shape[1])
for observation in data:
    s = online_empirical_cov(s=s, y=observation)
pprint(s)

This will return the running state, which includes the mean and cov

{'count': 10000,
 'cov': array([[0.38549265, 1.55198303, 0.73809891],
       [1.55198303, 9.41813338, 6.20756741],
       [0.73809891, 6.20756741, 4.79042088]]),
 'identity': array([[1., 0., 0.],
       [0., 1., 0.],
       [0., 0., 1.]]),
 'mean': array([2.19919374, 4.42189785, 1.52406171]),
 'n_dim': 3,
 'ones': array([1., 1., 1.]),
 'shape': (3, 3)}

To covert to corrcoef,

  from precision.covariance.util import cov_to_corrcoef
  pprint( cov_to_corrcoef(s['cov']) )

returns:

  array([[1.        , 0.81749783, 0.55345955],
   [0.81749783, 1.        , 0.92689045],
   [0.55345955, 0.92689045, 1.        ]])

See also

If you only need univariate, there is a really minimalise package momentum which avoids use of numpy.