Scaling differences in Brain Scores from Java and Matlab Contrast PLS analyses

GoogleCodeExporter commented 8 years ago

When viewing Java and Matlab contrast results side-by-side in Matlab
viewer, results look the same except there seem to be scaling differences
in Brain Scores: Java Brain Scores are about 50 times larger.

See Issue #83 for test file location details.

Original issue reported on code.google.com by anitaeo...@gmail.com on 31 Aug 2009 at 7:35

GoogleCodeExporter commented 8 years ago

Charisa has been working on PLS code and may have fixed this; she will have to 
verify.

Original comment by anitaeo...@gmail.com on 29 Jun 2010 at 7:49

GoogleCodeExporter commented 8 years ago

This is fixed and tested using multiple blocked and event-related datasets. 
Please see attached excel files comparing outputs for Non-rotated (contrast) 
PLS and Non-rotated Behavioral PLS (i.e., brain scores values). The Brain 
Scores for Java and Matlab contrast results are exactly the same.

The file 5cond_contrasts_BfMRIcontrast.txt no longer exist in thalia so was 
unable to replicate the analysis. 

Please see new test files located in /iris/grady_lab/omerdata/plsnpairs_testing/
-20yng16old_5cond_nonrotated_matlab_BfMRIresult.mat
-20yng16old_5cond_nonrotated_java_BfMRIresult.mat

Original comment by ng.char...@gmail.com on 21 Jul 2010 at 5:21

Attachments:

NonrotatedTaskPLS.xls
[Nonrotated Behav PLS.xls](https://storage.googleapis.com/google-code-attachments/plsnpairs/issue-84/comment-2/Nonrotated Behav PLS.xls)

GoogleCodeExporter commented 8 years ago

Original comment by anitaeo...@gmail.com on 2 Sep 2010 at 5:08

Changed state: Fixed

oleksiiparf / plsnpairs

Scaling differences in Brain Scores from Java and Matlab Contrast PLS analyses #84