[pymvpa] Sensitivity analysis with GNB

[Nick Oosterhof]

On 12 Dec 2014, at 16:13, Thomas Nickson <thomas.nickson at gmail.com> wrote:

>> >    # XXX Later come up with some
>> >    # could be a simple t-test maps using distributions
>> >    # per each class
>> 
>> yeah -- something like that might prove being useful, but probably not
>> too far from omnibus or ad-hoc Anovas which we have.  So that is why
>> they were never implemented
> You mean using ANOVAs to calculate the sensitivity? How do I do that? 

See e.g. http://www.pymvpa.org/generated/mvpa2.measures.anova.OneWayAnova.html

Since the GNB classifier works feature-wise (does not consider any covariance across voxels) the F value is a measure of how sensitive a feature is in discriminating between the classes.