[pymvpa] GroupClusterThreshold data / attributes

Tue Feb 28 19:19:47 UTC 2017

Hi,
> is there a way to use the AttributePermutator in a way that maintains the
same label shuffling across subjects? Even with cross-validation?
The easiest way to do this would be to set the same seed random seed for
each subject. AttributePermutator has a rng parmeter that does it.

BW,
Richard

On Tue, Feb 28, 2017 at 3:10 AM, Ryan Stolier <ryan.m.stolier at gmail.com>
wrote:

> Hi Richard,
>
> Thank you so much for the help, I am surprised you have this down to
> memory —as is, your example runs perfectly with the trained 'thr' object as
> all that is needed.
>
> One last question for you, or anyone who may know - is there a way to use
> the AttributePermutator in a way that maintains the same label shuffling
> across subjects? Even with cross-validation? I have read several sources
> saying this should be done for several reasons, eg, discussion here:
>
> http://mvpa.blogspot.com/2012/10/permuting-searchlight-maps-stelzer.html
>
> Thank you again for any help, this has been immense.
>
> Best,
> Ryan
>
>
> On Sun, Feb 26, 2017 at 7:02 AM <pkg-exppsy-pymvpa-request@
> lists.alioth.debian.org> wrote:
>
>> Send Pkg-ExpPsy-PyMVPA mailing list submissions to
>>         pkg-exppsy-pymvpa at lists.alioth.debian.org
>>
>> To subscribe or unsubscribe via the World Wide Web, visit
>>         http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/
>> pkg-exppsy-pymvpa
>>
>> or, via email, send a message with subject or body 'help' to
>>         pkg-exppsy-pymvpa-request at lists.alioth.debian.org
>>
>> You can reach the person managing the list at
>>         pkg-exppsy-pymvpa-owner at lists.alioth.debian.org
>>
>> When replying, please edit your Subject line so it is more specific
>> than "Re: Contents of Pkg-ExpPsy-PyMVPA digest..."
>>
>>
>> Today's Topics:
>>
>>    1. Re: GroupClusterThreshold data / attributes (Richard Dinga)
>>
>>
>> ----------------------------------------------------------------------
>>
>> Message: 1
>> Date: Sat, 25 Feb 2017 19:52:11 +0100
>> From: Richard Dinga <dinga92 at gmail.com>
>> To: Development and support of PyMVPA
>>         <pkg-exppsy-pymvpa at lists.alioth.debian.org>
>> Subject: Re: [pymvpa] GroupClusterThreshold data / attributes
>> Message-ID:
>>         <CABbjURCstmE8TgKQ8Qw=uYQSb52zm=sxDhGoNy1xR-
>> chAZTOSQ at mail.gmail.com>
>> Content-Type: text/plain; charset="utf-8"
>>
>> Hi,
>> I am sorry, my python is bit rusty by now, and I don't have a data set
>> ready to give you an reproducible example. However I think that once you
>> train your GroupClusterThreshold object, then the all cluster sizes from
>> your bootstraps should be in _null_cluster_sizes attribute. Then your
>> p-value for your arbitrary cluster is just a it's percentile in that
>> array.
>>
>> You can also use methods get_cluster_pvals or _transform_to_pvals from the
>> GroupClusterThreshold class to do it for you. Somehow :)
>> Actual code that computes the p-values for clusters is:
>> cluster_probs_raw = _transform_to_pvals( area, self
>> ._null_cluster_sizes.astype('float'))
>> so MAYBE this will work for you
>> cluster_probs_raw = GroupClusterThreshold._transform_to_pvals(
>> np.array([40. 60, 80]), thr._null_cluster_sizes.astype('float'))
>>
>> On Thu, Feb 9, 2017 at 10:40 PM, Ryan Stolier <ryan.m.stolier at gmail.com>
>> wrote:
>>
>> > Hi all,
>> >
>> > I recently implemented GroupClusterThreshold() based on help I found in
>> > this thread: https://lists.alioth.debian.org/pipermail/pkg-
>> > exppsy-pymvpa/2015q3/003202.html
>> >
>> > I was able to execute everything properly and get back results.
>> However, I
>> > am curious if there is a way to access data used along the way in the
>> > algorithm and figure out corrected significance levels of *any* size
>> > cluster, rather than only the clusters in my real (non-permuted) data.
>> Ie,
>> > can I access the table of cluster sizes from the 100000 bootstraps of
>> the
>> > permutation maps used to determine the significance levels of clusters
>> in
>> > my real data? Say I wanted to find out the hypothetical significance
>> level
>> > of clusters sized: 40, 60, 80 given everything besides the 'mean_map'
>> input
>> > in the email exchange in the listserv above. Does that make sense?
>> Also, I
>> > suppose the data used at steps 6/7 here: http://mvpa.blogspot.
>> > com/2012/10/permuting-searchlight-maps-stelzer.html
>> >
>> > Thank you so much for any assistance!
>> > Ryan
>> >
>> > _______________________________________________
>> > Pkg-ExpPsy-PyMVPA mailing list
>> > Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
>> > http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/
>> pkg-exppsy-pymvpa
>> >
>> -------------- next part --------------
>> An HTML attachment was scrubbed...
>> URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-
>> pymvpa/attachments/20170225/69ca5c95/attachment-0001.html>
>>
>> ------------------------------
>>
>> Subject: Digest Footer
>>
>> _______________________________________________
>> Pkg-ExpPsy-PyMVPA mailing list
>> Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
>> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
>>
>> ------------------------------
>>
>> End of Pkg-ExpPsy-PyMVPA Digest, Vol 107, Issue 6
>> *************************************************
>>
>
> _______________________________________________
> Pkg-ExpPsy-PyMVPA mailing list
> Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20170228/6da53bd7/attachment.html>