strucGlmer: Generalized linear mixed model with structured (co)variance...

In stevencarlislewalker/lme4ord: Generalized Linear Mixed Models with Flexible (Co)Variance Structures

Description

Fits a generalized linear mixed model with structured (co)variance terms by the following fully modularized steps:

strucParseFormula

parses the mixed model formula with structured terms

mkGeneralGlmerDevfun

constructs a generalized linear mixed model deviance function

bobyqa

optimizer the deviance function

mkStrucGlmer

constructs an object of strucGlmer-class

Usage

strucGlmer(formula, ...)

## Default S3 method:
strucGlmer(formula, ...)

## S3 method for class 'formula'
strucGlmer(formula, data, family, addArgs = list(),
  modularVerb = FALSE, optVerb = 0L, optMaxit = 10000, weights = NULL,
  offset = NULL, etastart = NULL, devfunOnly = FALSE,
  numModularSteps = 4L, parList = NULL, ...)

## S3 method for class 'strucParseFormula'
strucGlmer(formula, family, optVerb = 0L,
  optMaxit = 10000, weights = NULL, offset = NULL, etastart = NULL, ...)

## S3 method for class 'function'
strucGlmer(formula, optVerb = 0L, optMaxit = 10000,
  parsedForm, initPars, lower, upper, ...)

## S3 method for class 'strucGlmer'
update(object, formula, parList, data = NULL,
  addArgs = list(), optVerb = 0L, optMaxit = 10000, weights = NULL,
  offset = NULL, etastart = NULL, ...)

Arguments

formula	extended mixed model formula or strucParseFormula object
data,family,weights,offset,etastart,devfunOnly	see glmer
addArgs	list of additional arguments to pass to
modularVerb	should the modular steps be announced during the fitting process?
optVerb	verbose
optMaxit	maximum number of iterations for the nonlinear optimizers
numModularSteps	number of modular steps to perform (defaults to 4, which is all of them). 1 returns the strucParseFormula object, 2 the deviance function, 3 the optimizer output, and 4 a strucGlmer object.
parList	potential named list of initial parameters
parsedForm	strucParseFormula object
initPars	initial parameter values (required for strucGlmer.function method if missing parsedForm)
lower,upper	lower and upper bounds of parameter values (only useful for strucGlmer.function method)
...	further arguments to mkGeneralGlmerDevfun
object	strucGlmer object

Examples

cbpp$incidenceBySize <- with(cbpp, incidence/size)
gm <- strucGlmer(incidenceBySize ~ factAnal(0 + herd | period, nAxes = 1),
                 data = cbpp, family = binomial, weights = cbpp$size,
                 penLoads = mkPenLpNorm())

stevencarlislewalker/lme4ord documentation built on May 30, 2019, 4:43 p.m.