numpy.corrcoef not to be used optional variable ddof #4807

Open
crisrab opened this Issue Jun 13, 2014 · 2 comments

Projects

None yet

3 participants

@crisrab
crisrab commented Jun 13, 2014

while in the function numpy.cov it has a meaning to specify ddof, since numpy.corrcoef return the correlation coefficient that are the normalized covariance coefficient, ddof is irrelevant.
If I did not misunderstood the meaning of numpy.cov and numpy.corrcoef, I would suggest to remove the optional argument ddof in the function numpy.corrcoef.

@gustavla

I agree that it should be removed, since it's only a source of confusion. This also applies to bias.

It can make a minuscule numerical difference to change ddof, since it's still passed into cov and then normalized through its diagonal. However, I see no utility for this. In order to avoid having to pick ddof arbitrarily when calling cov, I suggest omitting it and thus letting it use its default value.

@rgommers
Member

@charris ddofwas added by you in a5b4a59, maybe you can comment?

Just removing the keyword will break some code, so shouldn't be done anyway. A clarification is preferable.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment