03-17-2010 05:49 AM
03-22-2010 09:08 AM
03-05-2016 09:18 AM - edited 03-05-2016 09:19 AM
I noticed the same problem and am sure the data is properly centered and scaled by the PLS procedure. I also used the centering/scaling output table from PROC PLS and applied it to the new data in PROC SCORE. See sample code attached to reproduce the issue. The results of the last proc means show a ~20% difference between the original factor generated by RRR and the factor generated by PROC SCORE (and the % difference is the same for all observations.) If any thoughts on what I may be missing or assuming incorrectly, please advice.