lava (version 1.5)

bootstrap.lvm: Calculate bootstrap estimates of a lvm object

Description

Draws non-parametric bootstrap samples

Usage

# S3 method for lvm
bootstrap(x,R=100,data,fun=NULL,control=list(),
                          p, parametric=FALSE, bollenstine=FALSE,
                          constraints=TRUE,sd=FALSE,silent=FALSE,
                          parallel=lava.options()$parallel,
                          mc.cores=NULL,
                          ...)
# S3 method for lvmfit
bootstrap(x,R=100,data=model.frame(x),
                             control=list(start=coef(x)),
                             p=coef(x), parametric=FALSE, bollenstine=FALSE,
                             estimator=x$estimator,weights=Weights(x),...)

Arguments

x

lvm-object.

R

Number of bootstrap samples

data

The data to resample from

fun

Optional function of the (bootstrapped) model-fit defining the statistic of interest

control

Options to the optimization routine

p

Parameter vector of the null model for the parametric bootstrap

parametric

If TRUE a parametric bootstrap is calculated. If FALSE a non-parametric (row-sampling) bootstrap is computed.

bollenstine

Bollen-Stine transformation (non-parametric bootstrap) for bootstrap hypothesis testing.

constraints

Logical indicating whether non-linear parameter constraints should be included in the bootstrap procedure

sd

Logical indicating whether standard error estimates should be included in the bootstrap procedure

silent

Suppress messages

parallel

If TRUE parallel backend will be used

mc.cores

Number of threads (if NULL foreach::foreach will be used, otherwise parallel::mclapply)

…

Additional arguments, e.g. choice of estimator.

estimator

String definining estimator, e.g. 'gaussian' (see estimator)

weights

Optional weights matrix used by estimator

Value

A bootstrap.lvm object.

See Also

confint.lvmfit

Examples

m <- lvm(y~x)
d <- sim(m,100)
e <- estimate(y~x, d)
 ## Reduce Ex.Timings
B <- bootstrap(e,R=50,parallel=FALSE)
B