Example? How to specify start values for theta and beta in lmer?

user_11677670 0
Viewed 745

I'm relatively new to R and still struggle with understanding some of the documentation. I'm working with a glmer analysis, and my goal is to specify starting values for both theta (intercept) and beta (fixed effects).

Can anyone please provide an example of what the code should look like for one beta and multiple betas?

I've looked at the glmer documentation, which says "start
a named list of starting values for the parameters in the model, or a numeric vector. A numeric start argument will be used as the starting value of theta. If start is a list, the theta element (a numeric vector) is used as the starting value for the first optimization step (default=1 for diagonal elements and 0 for off-diagonal elements of the lower Cholesky factor); the fitted value of theta from the first step, plus start[["fixef"]], are used as starting values for the second optimization step. If start has both fixef and theta elements, the first optimization step is skipped. For more details or finer control of optimization, see modular."

I have successfully specified the theta start value, but I do not know how to specify both theta and beta (or multiple betas).

Glucose1 <- glmer(GlucoseYN ~ 1 + (1|Subject), data=LongFormat, family = "binomial", glmerControl(optimizer = "bobyqa", optCtrl = list(maxfun = 10000000)))

getME(Glucose1, c("theta","beta"))
# theta = 3.276187, beta = -0.3515557

Glucose1b <- glmer(GlucoseYN ~ 1 + (1|Subject), data=LongFormat, family = "binomial", start = 3.276187, glmerControl(optimizer = "bobyqa", optCtrl = list(maxfun = 10000000)))
lme4 r
2 Answers

Here is one of the glmer examples with an addition of the start argument as it sounds like you wish to use it:

library(lme4)

glmer(
  cbind(incidence, size - incidence) ~ period + (1 | herd),
  data = cbpp,
  family = binomial,
  start = list(theta = 0.6, fixef = c(-1.4, -1, -1.1, -1.6))
)

So in your case:

Glucose1b <-
  glmer(
    GlucoseYN ~ 1 + (1 | Subject),
    data = LongFormat,
    family = "binomial",
    control = glmerControl(optimizer = "bobyqa", optCtrl = list(maxfun = 10000000)), 
    start = list(theta = 3.276187, fixef = -0.3515557)
  )
user_93932130

I've never used glmer but in non-linear regression the code is: 

Richards = nls(Hd ~ a/(1+b*exp(-c*Id)), data = Capacidade_Prod, start = list(a = 20, b = 4, c = 0.01))
user_116779140
Related