rankObjectWithRange: Rank a 'SetOfSchemes-class' or a 'SetOfTimePoints' object...

In microsamplingDesign: Finding Optimal Microsampling Designs for Non-Compartmental Pharmacokinetic Analysis

Description

Rank a SetOfSchemes-class or a SetOfTimePoints object using data generated per scneario defined by PkModelRange-class

Usage

rankObjectWithRange(object, pkModelRange, nSim,
  summaryFunctionOverScenarios = "max", directory = NULL, nCores = 1,
  seed = 123, ...)

Arguments

object	to be ranked
pkModelRange	see PkModelRange-class
nSim	number of samples dataset to generate per scenario (= combination of uncertain parameters)
summaryFunctionOverScenarios	function to summarize performance over different scenarios, defaults to max which corresponds to the min-max criterion
directory	directory to save models as .Rds objects, defaults to NULL when a temporary directory is made to save models and additional info on simulation settings, ranks, ...
nCores	number of cores used internally for ranking
seed	random seed reset when ranking on each directory ( for reproducibility ) , defaults to 123
...	additional parameters to pass to rankObject

Note

parallel computing at level of individual ranking and data generation

see rankObject for additional arguments, whem ranking a link{SetOfTimePoints-class} , nSubjectsPerScheme should be included

The same random seed is used when using parallel computations

Examples

  ## Not run:  # takes to much time for CRAN
   ## rank SetOfSchemes 
  setOfSchemesExample    <-  getExampleSetOfSchemes()
  pkModelRange           <-  getExamplePkModelRange()
  nSim                   <-  13 
  testDirectory1         <-  file.path( tempdir() , "test1" )
  dir.create( testDirectory1 )
  rankObjectWithRange( object = setOfSchemesExample , pkModelRange ,  nSim = 13  , 
    summaryFunctionOverScenarios = "max" ,
    directory  = testDirectory1 , varianceMeasure = "sd" , objective = getExampleObjective() 
    , nCores = 1 )

 
  ## rank set of timePoints 
  timePoints          <-  getExampleSetOfTimePoints( 0:10 )
  testDirectory2      <-  file.path( tempdir() , "test2" )
  dir.create( testDirectory2 )
  rankObjectWithRange( object = timePoints , pkModelRange ,  nSim = 13  ,
    summaryFunctionOverScenarios = "max" ,  directory  = testDirectory2 , nGrid = 20 , 
    nSamplesAvCurve = 25 , nSubjectsPerScheme = 3 
    , nCores = 1 )
  
  # remark : use larger number of simulation in realistic context
 
 ## clean up directories 
  unlink( testDirectory1 , recursive = TRUE )
  unlink( testDirectory2 , recursive = TRUE )
 
## End(Not run)

microsamplingDesign documentation built on Oct. 13, 2021, 5:10 p.m.