Learn R Programming
mixAK (version 5.8)

GLMM_MCMCifit: Initial (RE)ML fits for the GLMM_MCMC function

Description

This is a help function for GLMM_MCMC function. Besides initial (RE)ML fits, the function created variables derived from the design matrices.

THIS FUNCTION IS NOT TO BE CALLED BY ORDINARY USERS.

Usage

GLMM_MCMCifit(do.init, na.complete,
    y, dist, id, time, x, z, random.intercept,
    xempty, zempty, Rc, Rd, p, p_fi, q, q_ri, lalpha, dimb)

Value

A list with the following components (some of them not included if

do.init is FALSE):

Y

a list of length R with observations really used in fitting process (after removal of missing values)

ID

a list of length R with id's corresponding to Y

time

a vector time upon removal of missing values

x

a list resulting from the original argument x after removal of observations with some missing information additionaly, intercept column is added if fixed intercept included in the model

z

a list resulting from the original argument z after removal of observations with some missing information additionaly, intercept column is added if random intercept included in the model

I

number of subjects (grouped observations) in the original data (before removing NA's)

n

a list of length R, each component is a vector or length I (may contain zeros if some cluster disappears for particular response due to NA's)

Cn

vectorized n

sumCn

sum(Cn) = total number of observations

Cy_c

vector with continuous response to be passed to C++, equal to 0 if there is no continuous response

Cy_d

vector with discrete response to be passed to C++, equal to 0 if there is no discrete response

CX

vector containing X matrices (without ones for possible intercept) to be passed to C++, equal to 0 if there are no X matrices

CZ

vector containing Z matrices (without ones for possible intercept) to be passed to C++, equal to 0 if there are no Z matrices

iintcpt

data.frame(Est, SE) with estimated intercepts and their SE, R rows, row equal to (0, 0) if there is no fixed intercept for particular response

ifixef

a list of length R, each component is equal to 0 if there are no fixed effects for particular response, and is equal to data.frame(Est, SE) if there are fixed effects

isigma

vector of length R, equal to 0 for discrete response, equal to estimated residual standard deviation for continuous response

iEranef

a list of length R, each component is equal to 0 if there are no random effects for particular response, and is equal to data.frame(Est, SE) with estimated means of the random effects and their std. errors if there are random effects

iSDranef

a list of length R, each component is equal to 0 if there are no random effects for particular response, and is equal to a vector with estimated standard deviations of the random effects if there are random effects

ib

a list of length R, each component is equal to 0 if there are no random effects for particular response, and a matrix with EB estimates of random effects shifted by their estimated mean if there are random effects

is.intcpt

logical vector of length R

is.fixef

logical vector of length R

is.ranef

logical vector of length R

is.sigma

logical vector of length R

ibMat

matrix with initial values of random effects (EB estimates from (RE)ML fits)

ibMat2

matrix with alternative initial values of random effects

iEranefVec

vector with estimated means of random effects

iSDranefVec

vector with estimated standard deviations of random effects

iSEranefVec

vector with standard errors of estimated means of random effects

ialpha

vector with initial values of alpha's (including fixed intercepts)

ialpha2

vector with alternative initial values of alpha's (including fixed intercepts)

iSEalpha

vector with standard errors of estimated values of fixed effects

Arguments

do.init

logical value indicating whether initial (RE)ML fits should be done

na.complete

logical value. If TRUE then the function removes rows containing NA's from y, id, x, z whenever there is at least one missing value for arbitrary response. If FALSE then the missing values are removed response by response, i.e., different response variables may have different numbers of observations.

y

see output element y of GLMM_MCMCdata function

dist

see argumentdist of GLMM_MCMC function

id

see output element id of GLMM_MCMCdata function

time

see argument time of GLMM_longitClust

x

see output element x of GLMM_MCMCdata function

z

see output element z of GLMM_MCMCdata function

random.intercept

see output element random.intercept of GLMM_MCMCdata function

xempty

see output element xempty of GLMM_MCMCdata function

zempty

see output element zempty of GLMM_MCMCdata function

Rc

see output element Rc of GLMM_MCMCdata function

Rd

see output element Rd of GLMM_MCMCdata function

p

see output element p of GLMM_MCMCdata function

p_fi

see output element p_fi of GLMM_MCMCdata function

q

see output element q of GLMM_MCMCdata function

q_ri

see output element q_ri of GLMM_MCMCdata function

lalpha

see output element lalpha of GLMM_MCMCdata function

dimb

see output element dimb of GLMM_MCMCdata function

Author

Arnošt Komárek arnost.komarek@mff.cuni.cz

See Also

GLMM_MCMC.