Extract parameters from fitted model for use as initial values

This function extracts parameter values from a fitted model object in a form that can be directly provided as initial values for a new model fit.

Usage

# S3 method for class 'galamm'
extract_optim_parameters(object)

Arguments

object

Object of class galamm returned from galamm.

Value

A list object containing the following elements:

• theta Numerical vector of variance components, i.e., entries of the lower Cholesky form of the covariance matrix of random effects.

• beta Fixed regression coefficients.

• lambda Factor loadings.

• weights Weights for heteroscedastic residuals.

See also

Other optimization functions: galamm_control()

Examples

# Fit linear mixed model with heteroscedastic residuals
mod <- galamm(
  formula = y ~ x + (1 | id),
  weights = ~ (1 | item),
  data = hsced
)

# Extract parameters
start <- extract_optim_parameters(mod)

# Fit again using the Nelder-Mead algorithm, using start as initial values:
mod_nm <- galamm(
  formula = y ~ x + (1 | id),
  weights = ~ (1 | item),
  data = hsced,
  start = start,
  control = galamm_control(method = "Nelder-Mead")
)