[pymvpa] SMLR weights

[Michael Hanke]

On Wed, Jan 21, 2009 at 12:49:38PM -0800, Daqiang Sun wrote:
> Dear all,
> 
> I just started learning PyMVPA and ran some structural imaging data of patients and controls.
Great. Tell us if you get some nice results!

> To get SMLR weights, I tried both SMLRWeights(SMLR()) and
> SMLR().getSensitivityAnalyzer(). Although they give the same selected
> voxels, but the values are different. Which one should I use?
Any way will compute exactly the same algorithm. However, the SMLR
training process contains random elements so you cannot expect that
exactly _identical_ weights will result from training on the same
dataset. That should even be the case if you simple compute the weights
twice with a single method.

> I
> assumed that the dot product of weights and selected voxel values
> (gray matter density in my case) would give an idea about how the two
> groups could be separated, but it looks not the case. Although the
> prediction accuracy is pretty good (85%), the dot products for the two
> groups are largely overlapped. I guess I just misunderstood it. Could
> you give me some idea about what the weights are about? 
The SMLR classifier provides these values for you -- no need to compute
them manually. Please take a look at the 'pylab_2d' example

	http://www.pymvpa.org/ex/pylab_2d.html

which shows how to extract these 'decision values' from a range of
classifiers.


Cheers,

Michael


-- 
GPG key:  1024D/3144BE0F Michael Hanke
http://apsy.gse.uni-magdeburg.de/hanke
ICQ: 48230050