Closed namtien0312 closed 2 years ago
Hi @namtien0312
You cannot use norm(cp*prop.sd)
instead use prop(prop.sd)
For the lognormal case you can use
cp ~ lnorm(expo.sd)
In this case, you can compare the objective function from the add
and prop
error models since they are on the same scale.
Ohh, Thank Dr. Fidler for taking the time to help!
Tien Nguyen.
Dear
nlmixr
team,I found that additive RUV
cp ~ add(add.sd)
is equivalent tocp ~ norm(add.sd)
in userguide book!How can I define in this way for proportional RUV, I tried
cp ~ norm(cp*prop.sd)
but it doesn't work, with error like that:Error in nlmixrUIModel(.model, ini, fun) : residual distribution parameter(s) estimates were not found in ini block
. Code is:And, with exponential RUV, how can I define it in the model. Is that
log(cp) ~ norm(expo.sd)
?Thank you for the explanation! Tien Nguyen.