samplr: generates a Latin hypercube based on inputs provided in an...

In alexmitrani/humblr: Provides a few humble tools for R.

generates a Latin hypercube based on inputs provided in an Excel spreadsheet.

Description

facilitates the generation of a Latin hypercube for a set of variables based on inputs specified in a spreadsheet table.

Usage

samplr(
  myfilename = NULL,
  mytestspervariable = 10,
  myseed = 12345L,
  mymaxsweeps = 4,
  myeps = 0.01,
  mydup = 5,
  mypop = 1000,
  mygen = 8,
  mypmut = 0.1,
  mygraphsize = 1000,
  mypch = 19,
  mycol = "blue",
  mycex = 0.5
)

Arguments

myfilename	is the name of the input Excel file
mytestspervariable	the number of tests per variable in the input file
myseed	the random number seed to be used
mymaxsweeps	see lhs package documentation
myeps	see lhs package documentation
mydup	see lhs package documentation
mypop	see lhs package documentation
mygen	see lhs package documentation
mypmut	see lhs package documentation
mygraphsize	pixel dimension to be used for scatter graph matrices. see "width" and "height" in png grDevices.
mypch	see "pch" in par graphics
mycol	see "col" in par graphics
mycex	see "cex" in par graphics

Details

The spreadsheet should have only one sheet with the following columns and one row for each variable to be simulated:

• Index (numeric)

• Variable (character)

• codename (character)

• Base (numeric)

• Minimum (numeric)

• Mode (numeric)

• Maximum (numeric)

• Distribution (character) - Uniform, Triangular, or PERT.

• Shape (numeric) - only needed if distribution is "PERT"

Six types of optimisation algorithm are tested:

• randomLHS

• optimumLHS

• maximinLHS

• improvedLHS

• geneticsLHS - genetic algorithm with "S" criterium

• geneticmLHS - genetic algorithm with "Maximin" criterium

The algorithm that minimises the maximum correlation ("max_corr") is selected to produce the recommended Latin hypercube. The details of the selected algorithm and the other algorithms tested are added to the sheet "optimisation_summary" of the results workbook.

Examples

myinputfile <- system.file("extdata", "risk_variable_distributions.xlsx", package = "humblr")
mytest1 <- samplr(myinputfile, mytestspervariable = 10, myseed = 12345L, mymaxsweeps = 4, myeps = 0.01, mydup = 5, mypop = 1000, mygen = 8, mypmut = 0.1, mygraphsize = 1000, mypch = 19, mycol = "blue", mycex = 0.5)
mytest1

alexmitrani/humblr documentation built on April 4, 2022, 8:29 a.m.