PLS - Important variable selection - Unscrambler
Submitted by janskvaril on 9 September 2015 - 3:32pm
Forums:
Dear Sir or Madame,
I would like to ask if there is a method how to identify important X-variables for a prediction of Y-variable(s) in Unscrambler. I'm using PLS not only for spectroscopic calibration purposes and my models are having many components (i.e. each of the first 4-5 PCs explain 15-20%) and therefore the PC1/PC2 plot is not enough to understand which are the most important variables.
Of course I know of Martens uncertainty test and I can also take a look at the BW-weighted regression however, please are there any other methods how to identify important variables in PLS model within Unscrambler?
Thank you
With kind regards
Jan Skvaril