Reduce variates in (s)PLS-DA

[Kim-Anh Le Cao]

Florian has highlighted some discrepancies in the PLS vs sPLS code regarding the scaling of the variates.

According to Tenehaus (p204), the regression mode does not scale the t_h and v_h. In the current mixOmics code, the sPLS scale these variates, but the PLS does not.

Florian is testing sPLS with no scaling and see if the sPLS code should be amended.

[Kim-Anh Le Cao]

Feedback from Florian: the results might affect the (s)PLS-DA prediction

[Kim-Anh Le Cao]

done