| bootstrap.lvm {lava} | R Documentation |
Draws non-parametric bootstrap samples
## S3 method for class 'lvm'
bootstrap(x,R=100,data,fun=NULL,control=list(),
p, parametric=FALSE, bollenstine=FALSE,
constraints=TRUE,sd=FALSE,silent=FALSE,
parallel=lava.options()$parallel,
mc.cores=NULL,
...)
## S3 method for class 'lvmfit'
bootstrap(x,R=100,data=model.frame(x),
control=list(start=coef(x)),
p=coef(x), parametric=FALSE, bollenstine=FALSE,
estimator=x$estimator,weight=Weight(x),...)
x |
|
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) |
estimator |
String definining estimator, e.g. 'gaussian' (see
|
weight |
Optional weight matrix used by |
... |
Additional arguments, e.g. choice of estimator. |
A bootstrap.lvm object.
Klaus K. Holst
m <- lvm(y~x) d <- sim(m,100) e <- estimate(y~x, d) ## Reduce Ex.Timings B <- bootstrap(e,R=50,parallel=FALSE) B