16 Multiblock Canonical Correlation and overals()

16.1 Equations

16.2 Examples

16.2.1 Thirteen Personality Scales

This is the same example as before, but now we group the five scales from the Eysenck Personality Inventory and the five from the Big Five inventory into blocks. The remaining three variables define three separate blocks. No copies are used, and we use monotone cubic splines with the interior knots at the quartiles.

epi_knots <- lapply (epi, function (x) fivenum (x)[2:4])
epi_degrees <- rep (3, 13)
epi_blocks <- c(1,1,1,1,1,2,2,2,2,2,3,4,5)
h <- overals(epi, epi_blocks, epi_copies, epi_knots, epi_degrees, epi_ordinal)
In 191 iterations we find minimum loss 0.030226. The object scores are in figure 25 and the transformation plots in figure 26.
Figure 25: Personality Scales, Multi-Set, Objects Scores


Figure 26: Personality Scales, Multi-Set, , Transformations