[pymvpa] Suspicious results: problem solved!

[Yaroslav Halchenko]

with pymvpa 0.4 using pynifti as the backend to access NIFTIs --
scl_slope is not used at all.  And due to detrending/zscoring it is
usually irrelevant (unless you believe in percent-signal-changes, when
scl_interc and scl_slope would become relevant), that is why we never
ran into such situation before.

in 0.6, we use nibabel and that one is meticulous enough to apply and
de-apply slope and intercept upon load/save operations.

> One last question: do I need to set the scl_slope back to 1,2222
> before proceeding with the next dataset (if I understand correct this
> value is also used during the importing of the data?)

well, new dataset will come from another nifti file I guess, so it will
be completely different instance assigned to 'dataset' variable, so it
will get the corresponding slope from the file.  You would need to reset
it to 1.0 again before map2nifti if you are going to store the
accuracies etc.

-- 
=------------------------------------------------------------------=
Keep in touch                                     www.onerussian.com
Yaroslav Halchenko                 www.ohloh.net/accounts/yarikoptic