Page 1 of 1

MSLiveDead Issue

PostPosted: Tue Jan 16, 2024 8:12 pm
by wradigan
Hi,

Jeff, I've emailed you the input file "monthly.txt".

I'm getting the messages "Note: only 3 parameters counted of 4 specified parameters" and "AICc and parameter count have been adjusted upward" in R and "Attempted ordering of parameters by estimatibility: 2 1 3 4" and "Beta number 4 is a singular value." in the output file when I run my below code:

Code: Select all
library(RMark)

try2=import.chdata("monthly.txt",header=FALSE,
                   field.names=c("ch","freq","Sex","Species"),field.types=c("n","n","n"))

try2$Sex <- as.factor(try2$Sex)
try2$Species <- as.factor(try2$Species)

wall.processed=process.data(try2, model="MSLiveDead", groups = c("Sex","Species"),strata.labels=c("A","B","C","D"))

wall.ddl=make.design.data(wall.processed)
str(wall.ddl)

table(wall.ddl$Psi[,c("stratum","tostratum")])

r.dot=list(formula=~1)
p.dot=list(formula=~1)
S.dot=list(formula=~1)
Psi.dot=list(formula=~1)

r.stratum=list(formula=~1+stratum)
p.stratum=list(formula=~1+stratum)
S.stratum=list(formula=~1+stratum)
Psi.stratum=list(formula=~stratum)

model.list1=create.model.list("MSLiveDead")
mstrata.telem.resultsNEW=mark.wrapper(model.list1,data=wall.processed,ddl=wall.ddl)

return(mstrata.telem.resultsNEW)
mstrata.telem.resultsNEW



I've found similar posts on the forum but I don't understand why I'm hung up.

Re: MSLiveDead Issue

PostPosted: Tue Jan 16, 2024 9:52 pm
by jlaake
Use ~-1+stratum or ~stratum

For Psi you will want ~-1+stratum:tostratum

The message is because a parameter is at a boundary or because the model is over-parameterized which means you have specified more parameters than can be fitted. Likely it is the 1+stratum but without running it i can't be certain.

You can use model.matrix with the formula and the design data for a parameter to see the design matrix. That is created. Please read the workshop notes. A link is given every time you type library (RMark).