Skip to main content
Skip table of contents

getModelBuildingResults

Get the results of automatic covariate model building or automatic statistical model building. The exact details of what is returned depend on the strategy used when running the model building.

Usage

R
getModelBuildingResults()

Value

A list containing the results of model building with one element for each model run. For all strategies, each list item contains the following:

  • LL: result of -2*Log-Likelihood

  • BICc: modified BIC

  • individualModels: (data.frame of logical values) where the rows are the parameters in the model and the columns are the covariates, and the TRUE/FALSE value indicates if a covariate is used for that parameter

COSSAC returns two additional fields:

  • tested: (vector) which parameter-covariate pair was testing in this run with respect to the previous model, where the first element is the individual model parameter and the second one is the covariate

  • bestModel: (logical) whether this model is the best model amongst all the tested models according to the chosen criterion

SAMBA returns the error model and covariance model information if they exist:

  • errorModels: (data.frame) onservation model where each row specifies the observation id and the error model chosen

  • covarianceModels: list with one element for each variability level consisting of two elements: level, specifying the variability level, and correlations, a list of the chosen correlations between individual model parameters in the groups element

See also

getModelBuildingSettings for a description of settings
runModelBuilding to run model building

Examples

R
initializeLixoftConnectors("monolix")
project_file <- file.path(getDemoPath(), "1.creating_and_using_models", "1.1.libraries_of_models", "theophylline_project.mlxtran")
loadProject(project_file)
runModelBuilding(strategy = "samba")
res <- getModelBuildingResults()
JavaScript errors detected

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

If this problem persists, please contact our support.