Feature Request: Population Covariance Calculation with ddof parameter for DataFrame.cov() (original) (raw)

I propose adding a bias parameter in DataFrame.cov(). Currently, Pandas v0.25.1 does not have such a parameter.

The default would be bias=False, which would compute sample covariance with a N-1 factor in the denominator. When bias=True, this would compute population covariance with N in the denominator.

This follows the same semantics as the numpy.cov() function. Since DataFrame.cov() uses numpy.cov(), I hope this would not be too hard to implement.