If this is for the Science Fair, you'd probably have to have a vet
supervising the experiment, to make sure the ferrets aren't harmed.  (We
went throught this with my kid's project one year.  Not impossible, but
often hard to manage.)
 
Rather than diet, you might consider something that doesn't potentially
affect the ferrets' well-being at all, if you really want to do something
using animals.  (Not that you would do anything harmful, but judges tend to
be critical of animal experimentation.  There were some true horror stories
back before Science Fair adopted stricter regulations.) You might consider
something on ferret learning abilities.  When I was in 10th grade (a
zillion years ago), I did an experiment on the social learning abilities of
guinea pigs.  I found that one guinea pig was unlikely to ever get to the
end of a simple maze.  Two might eventually get there, and four almost
always got there in less than ten minutes.  My brother's pet rat, in
contrast, hopped over the maze wall and went straight to the treat, in
about 11 seconds the first time.
 
How about something that doesn't involve working with specific ferrets,
such as a survey of ferret coloring and incidence of adrenal disease.  (I
doubt there's a correlation, but it sure would be interesting if there
was!) Or track and map ferret distemper, and correlate it to either the
sources of the ferrets or the shipping routes.  Epidemiology would avoid
the difficulties of live animal research and could generate some useful
information.
[Posted in FML issue 2853]