cond.ranefs {gmvjoint} | R Documentation |
Obtain conditional distribution of the random effects
Description
Obtain the conditional distribution of the random effects of a joint
model
fit. This is achieved by a Metropolis scheme. Approximate normality across random effects is
expected, and could be useful in diagnosing potential issues surrounding model fits.
Usage
cond.ranefs(fit, burnin = 500L, N = 3500L, tune = 2)
Arguments
fit |
a joint model fit by the |
burnin |
Number of burn-in iterations to discard, defaults to 500. |
N |
Number of MC iterations to carry out post burn-in, defaults to 3500. |
tune |
Tuning parameter, problem-specific, defaults to 2. |
Value
A list of class cond.b.joint
containing:
- walks
A list of length
n
containing the history ofpost burn-in.
- acceptance
A numeric vector containing the acceptance rate for each sampled subject.
- M
The ModelInfo list from
joint
. Used by S3 methods for classcond.b.joint
.- bhats
Posterior estimates at MLEs for the random effects. Same as
ranef(joint)
.- Sigmahats
The covariances of
bhats
.- D
The MLE estimate for the variance-covariance matrix of random effects from
fit
.- q
Dimension of random effects.
- K
Number of responses.
- qnames
The names of the random effects as determined by call to
joint
.- burnin
The amount of burn-in used.
- N
Number of MC iterations.
- tune
tuning parameter used
- nobs
The number of observations for each subject for each response.
- elapsed.time
Time taken for
cond.ranefs
to complete.
See Also
Examples
dat <- simData()$data
long.formulas <- list(Y.1 ~ time + cont + bin + (1 + time|id),
Y.2 ~ time + cont + bin + (1 + time|id))
surv.formula <- Surv(survtime, status) ~ bin
fit <- joint(long.formulas, surv.formula, dat, list("gaussian","gaussian"))
cond.b <- cond.ranefs(fit, burnin = 50L, N = 1000, tune = 2)
cond.b
plot(cond.b) # Overall
plot(cond.b, id = 1) # Plot the first subject (see plot.cond.b.joint).