I am fitting a brmultinom now (instead of previous multinom), but I can not get the VIF or dredge function working on the new model. Any advise?
my code is
library(MuMIn)
options(na.action = na.fail)
dat$Sex <- as.factor(dat$Sex)
dat$season <- as.factor(dat$season)
dat$Data.origin <- as.factor(dat$Data.origin)
str(dat$Length)
str(dat$VBT)
#
model2 <- brmultinom(prey_new ~
Sex +
season +
Data.origin +
Length +
VBT,
data = dat,type="AS_mean",maxit=999)
where prey_new is a matrix
summary(model2)
dredge_model_multi <- dredge(model2, rank = "AICc")
and then it starts doing it (fixed term is intercept), but then gives an error Fixed term is "(Intercept)" Error in cf[, 2L] : subscript out of bounds
options(na.action = na.omit)
vif(model2, na.rm = T)
Error in if (names(coefficients(mod)[1]) == "(Intercept)") { : argument is of length zero
Previously on the multinom (nnet) I coudl also run a pseudor2 which doesnt run anymore now:
library("DescTools")
# Calculate the R Square
PseudoR2(model2, which = c("CoxSnell","Nagelkerke","McFadden"))
Error in match.arg(type) : 'arg' should be one of “class”, “probs”
Any experiences with this/ tips on how to solve the issues?
So, I fixed it by extracting the model calls through dredge with the multinom model but with evaluate=F and then looping all possible formulas with brmultinom and extracted AICc.