[Numpy-discussion] performance matrix multiplication vs. matlab

Keith Goodman kwgoodman at gmail.com
Mon Jun 8 09:28:06 EDT 2009


On Mon, Jun 8, 2009 at 6:17 AM, Gael Varoquaux
<gael.varoquaux at normalesup.org> wrote:
> On Mon, Jun 08, 2009 at 09:02:12AM -0400, josef.pktd at gmail.com wrote:
>> whats the actual shape of the array/data you run your PCA on.
>
> 50 000 dimensions, 820 datapoints.

Have you tried shuffling each time series, performing PCA, looking at
the magnitude of the largest eigenvalue, then repeating many times?
That will give you an idea of how large the noise can be. Then you can
see how many eigenvectors of the unshuffled data have eigenvalues
greater than the noise. It would be kind of the empirical approach to
random matrix theory.



More information about the NumPy-Discussion mailing list