Dear list,
I have a seemingly simple problem with plotting predictions from nlme for
which I have not been able to find an answer in the archives nor do I see
any mention of this in Pinheiro & Bates 2000. I am using the following model
of chick growth for seven different species, with season, nest and
individual chick as random effects (all of which are categorical).
chickg<-groupedData(nmass~age|fspecies, chick, order.groups=F)
logistic<-deriv(~A/(1+(((A-I)/I)*exp(-age*K))),
c("A","I","K"),
function(age,A,I,K) {})
ab<-nlme(nmass~logistic(age,A,I,K), chickg, fixed=A+I+K~fspecies,
random=A+K~1|fseason/fnest/fchick, start=fab6)
I would like a separate plot for each species (fspecies). As of yet, I have
only been able to plot by fseason using
plot(augPred(ab, level=0))
or by fchick when no level is set. getGroups(ab) returns 2295 levels, one
for each individual. How does one specify that the fixed effect should be
used as the grouping factor?
Thank you,
M
[[alternative HTML version deleted]]