hi,
i'm looking for a smart way of distributing items to different groups
according to their correlation(s). the correlation of items within one
group should be minimal, whereas (canonical) correlation between the
blocks should be maximal.
example: you have 20 items, which are to be split into 4 blocks with 5
items in each block. (of course, the numbers are to be interchangable).
so far, 2 (problematic) solutions were thought of:
1) try all different combinations of blocks of items and look for the
maximum canonical correlation between the blocks. problems are, that the
correlation of items within one block are only regarded implicitely and,
more sadly, this approach is very time-consuming as the number of items
increases.
2) an approximation could be to use the results of a factor analysis to
distribute items from one factor to different blocks. however, this is a
rather quick and dirty solution.
has anybody else encountered a similar problem before or can think of an
approach to handle this?
any help is more than welcome.
best regards
andreas
[[alternative HTML version deleted]]
______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
