[SciPy-user] Correlation coefficients

Bruce Southey bsouthey@gmail....
Wed Dec 12 08:42:12 CST 2007


Hi,
What exactly do you want as an outcome? In particular, is there a
simpler outcome such as only the covariance?

Are you correlating one array with each array of the pool (or against
the pool average)?
Or correlating each array in the pool with each other?

You may only need the covariance or crossproduct if the variance of
each array are essentially the same or covariance is sufficient. In
that case a array by array multiplication (pool as a multi-dimensional
array) will be sufficient. If not, create your own correlation
function where you store the mean and variance of each array the pool
so you only compute it once.


Regards
Bruce






On Dec 11, 2007 8:28 PM, Roger Herikstad <roger.herikstad@gmail.com> wrote:
> Hi all,
>  I was wondering if there is an efficient way of determining the degree of
> correlation between an array and and pool of reference arrays (> 10,000).
> I'm using corrcoef when the pool is small ( <1,000), but I have cases where
> I need to compare against more than 10,000 arrays, in which case corrcoef is
> slow. Any suggestions? Thanks!
>
> ~ Roger
> _______________________________________________
> SciPy-user mailing list
> SciPy-user@scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>


More information about the SciPy-user mailing list