I am performing a PLS regression between two sets of variables. The first
one are landmark coordinates of Procustes aligned specimens. The second one
are ecological variables. These ecological variables have different units.
Can I input these ecological variables in block 2 even if they have
different units? Or should I input Z-scores of these variables, for
example? I am asking becasue I have tried doing it with the raw values and
the results are that the loadings for one of the variables in PLS1 (which
explains almost all the variation) is almost 1, and the loadings for the
rest of the variables are almost 0. I interpret this as: shape is
associated to this block of variables basically because it is associated to
that ONE variable. But I am wondering if I am getting those results because
that ONE variable has huge values (and so a huge variation range) whereas
the rest have very low values (because of the units).
Also, I take this opportunity to ask something I am not sure of: do the
variables within one block have to be independent from each other?
Thanks very much!
MORPHMET may be accessed via its webpage at http://www.morphometrics.org
You received this message because you are subscribed to the Google Groups
To unsubscribe from this group and stop receiving emails from it, send an email