[pymvpa] ideas for contribution
debian at onerussian.com
Mon Mar 14 15:45:20 UTC 2016
On Mon, 14 Mar 2016, H. Y. Chan wrote:
> I've just finished recently porting the Matlab codes of a non-parametric
> ICA procedure (see http://dx.doi.org/10.1016/j.neuron.2015.11.035) to
> Python and wonder if you'd like to fit that into your package.
sure -- sounds interesting. I am a bit wary about the new numerical
methods though which were introduced without proper independent (of
specific application) analysis and comparison to other existing ones
(e.g. to FastICA they reference). But I guess it would be only of
benefit (at least for replication ;)) to include this one in PyMVPA. I
guess it would need to come as a Mapper akin to SVDMapper
> Also, I'd suggest adding Kendall tau correlation as additional option in
> RSA (jerry-rigged it myself but it'd be nice to incorporate it
I guess we should just use scipy/stats/stats.py:kendaltau implementation
and indeed expose it in RSA... should just rank order data first, if I got it
> Let me know how I can contribute.
Sure! e.g. you could start with a PR for that ICA mapper. The only
thing is that we would need tests to verify its correct (according to
some assumptions ;) ) operation
> Again, thanks for all the efforts that go to the toolbox!
Thank you for seeking contribution to it!
Yaroslav O. Halchenko
Center for Open Neuroscience http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834 Fax: +1 (603) 646-1419
More information about the Pkg-ExpPsy-PyMVPA