CODEWITHSUNDEEP
pythoncovariancerobust
OverFlow Police
OverFlow Police

Reputation: 861

Why the output of covariance functions in Python are too different?

I am using three functions of python to compute covariance of the same input, the outputs are drastically different. Does anyone have experience and know which one works best? (what are the differences?)

The functions I use are

sklearn.covariance.empirical_covariance(.)
MinCovDet().fit(.)
np.cov(.)

Any insight is appreciated.

The sklearn.covariance.empirical_covariance(.) gives me the straightforward 

cov = (1/N) * M.transpose * M

Upvotes: 1

Views: 193

Answers (1)

Michael B
Michael B

Reputation: 578

Agree with Joseph Hansen, more specificity will be helpful for a thorough answer. Quickly, I believe that sklearn.covariance calculates covariance for the population. Whereas, by default numpy.cov calculates the sample covariance. To obtain the population covariance you can specify normalization by the total N samples like this:

Covariance = numpy.cov(a, b, bias=True)[0][1]

Upvotes: 1

Related Questions