[pymvpa] Accuracy for distinguishing each class against all others

[jian hou]

Hi Brain,
Thank you very much for your reply.
I got the confusion matrix. How could I get the p value for each class?  Can I  just compare the classification accuracy for each class with the cvte.null_dist.ca.dist_samples?
Best,
Jian Hou








At 2016-04-11 02:53:13, "Brian Murphy" <brian.murphy at qub.ac.uk> wrote:
>Hi Jian,
>
>this tutorial page includes an example of multiclass analysis, and down
>the end it shows how to get the confusion matrix out of a (in this case
>cross-validated) classifier with cvte.ca.stats.matrix. Note that you
>have to tell your classifier to retain that information, with
>enable_ca=['stats'],
>
>best of luck,
>
>Brian
>
>
>
>
>
>On Sun, 2016-04-10 at 16:24 +0800, jian hou wrote:
>> Hi all,
>> I'd like to run a permutation test for a multi-class classification
>> and I'm interested in the classification accuracy of each class rather
>> than the overall accuracy. 
>> Is there a way to get the null distribution of the accuracy of each
>> class?
>> Any suggestions? Thanks in advance.
>> Best,
>> Jian Hou
>> 
>> 
>>  
>> 
>> _______________________________________________
>> 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
>
>-- 
>Brian Murphy
>Lecturer (Assistant Professor)
>Knowledge and Data Engineering
>School of EE Engineering and Computer Science
>Queen's University Belfast
>
>
>_______________________________________________
>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/20160411/f6f9a76d/attachment.html>