Manually generate sPLS component scores

Support
1

Hello!

I am trying to manually generate sPLS component scores for each participant by multiplying the values of the selected variables in the input X dataset by the loadings output from the sPLS function, and then summing the weighted values. However, the results don’t match the expected output from the sPLS function, even if I standardize the input X dataset so that each variable has a mean of 0 and a variance of 1.

Could you please advise on how I can recreate component scores for each participant manually using the input X dataset and the loading values for each variable from sPLS? A reviewer has asked me to generate component scores in a larger sample, some of whom are missing values for Y, to test if my results generalize.

Thanks!
Fran

1 Like
2

Hi @fquerdasi,

If you have a larger sample, some of which are missing Y, we would recommend making use of the mixOmics function predict(). This function is a great way to assess the generalisability of your model, you can read more about it on our website. Manually generating the scores might be tricky due to some details of the model building with mixOmics like scaling/centering/deflation/etc.

Hope that helps!
Eva