Skip to main content
Skip table of contents

getBootstrapSettings

Get the settings that will be used during the run of bootstrap.

Usage

R
getBootstrapSettings()

Value

The list of settings

  • nbRuns [optional] (integer) number of bootstrap replicates (default=200)

  • method [optional] (character) sampling method: "parametric" or "nonparametric" (default: nonparametric)

  • initialValues [optional] (character) initial values used in the boostrap runs for the estimation of the population parameters: "initial" or "final (default: "initial")

  • cens [optional] (list) if method="nonparametric" and there are censored observations in the dataset. A list, or a list of lists with elements obsid, type ("left,"right" or "interval") and limit (single value or vector of two values). Ex: list(list(obsid="y1",type="left", limit=0.1), list(obsid="y2", type="interval", limit=c(0.2,10))

  • tasks [optional] (list of character) tasks to perform in bootstrap runs in addition to population parameter estimation. Available tasks: "standardErrorEstimation", "logLikelihoodEstimation" (default=list())

  • useLin [optional] (logical) calculation method to estimate standard errors and log-likelihood (default = FALSE). If TRUE, they are estimated via linearization. If FALSE, standard errors are estimated via stochastic approximation and log-likelihood via importance sampling.

  • sampleSize [optional] (integer) the number of individuals in each bootstrap data set (default value is the number of individuals in the original data set).

  • covStrat [optional] (list of character) one or several categorical covariates of the project. The original distribution of this covariate is maintained in each resampled data set if covStrat is defined (default=list()). Notice that if the categorical covariate is varying within the subject (in case of occasions), it will not be taken into account.

  • level [optional] (numeric) level of the bootstrap confidence intervals (default = 0.95)

  • saveResultsFolders [optional] (logical) to choose if bootstrap projects results folders should be saved or deleted (default = FALSE)

  • saveDatasets [optional] (logical) to choose if bootstrap datasets and mlxtran files (Monolix project) should be saved or deleted (default = FALSE)

  • replaceFailedRuns [optional] (logical) to choose if bootstrap runs with failed convergence (maximum number of iterations reached before the autostop criterion) should be replaced by new runs (default=FALSE)

  • maxNbFailedRuns [optional] (integer) if replaceFailedRuns=TRUE, maximum number of runs with failed convergence that can be replaced before bootstrap is stopped (default=20)

See also

Examples

R
if (FALSE) {
# run parametric bootstrap with 100 runs
set = getBootstrapSettings()
set$nbRuns = 100
set$method = "parametric"
runBootstrap(set)

# run nonparametric bootstrap with 500 runs, stratified resampling by STUDY and DOSEGROUP categorical covariates, standard errors and log-likelihood estimated via linearization, and bootstrap datasets and mlxtran files saved in the results.
runBootstrap(nbRuns=500, method="nonparametric", covStrat=list("STUDY", "DOSEGROUP"), tasks=list("standardErrorEstimation", "logLikelihoodEstimation"), useLin=FALSE, saveDatasets=TRUE)
}
JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.