[R] Help with mixOmics rcc-tune

[Kim O]

Dear,


I am using the mixOmics rCCA package, but I am encountering a problem that
I cannot solve with the documentation only:


*How should I interpret the "optimal score" (opt.score) returned by
tune.rcc? *


It does not say what this score actually represent? My best guess would be
the average canonical correlations on the test set, but I am not sure?


In my specific application, I get canonical correlation of about 0.8 on
training data but a CV-score of only 0.20.... I don't know whether this is
good or bad and how to interpret this.


Can you provide information on how exactly this is calculated and how to
interpret it?


Thank you so much. Would mean so much to me!

Best,

Kim Rants

	[[alternative HTML version deleted]]