dsc: Obtain the best configuration to meet the objectives...

In Antoine-Masse/KefiR: KefiR Package

Obtain the best configuration to meet the objectives determined by one or more linear models.

Description

Obtain the best configuration to meet the objectives determined by one or more linear models.

Usage

dsc(
  data,
  reg,
  Y = c(),
  ymin = c(),
  ymax = c(),
  pop = iter/20,
  iter = 4000,
  wash = pop/2,
  plot = T,
  verbose = F,
  save = F,
  file = "file.html"
)

Arguments

data	A data.frame with X(s) and Y(s).
reg	A linear model or a list of linear models.
Y	Values that we want to achieve for the different Y predicted using the model (s).
ymin	List of minimum values tolerated for the different Y.
ymax	List of maximum values tolerated for the different Y.
pop	Population of parameters which will cross randomly to generate better parameters.
iter	Number of iterations in the scalable approach (should ideally be much greater than the popupulation (pop) of settings.
wash	The maximum number of desired settings.
plot	If TRUE, displays interactive parallel coordinates (plot_ly) to identify the best possible settings.
verbose	If TRUE, gives information about the analysis.
save	For saving the graph (html format)
file	Name of the html page "xxx.html"

Value

A dataframe containing all the selected settings sorted from best (top) to worst (bottom).

Examples

data(mtcars)
colnames(mtcars)
myreg1 <- evolreg(mtcars,"mpg")
myreg2 <- evolreg(mtcars,"cyl")
reg <- list()
reg[[1]] <- myreg1
reg[[2]] <- myreg2
output <- dsc(mtcars,reg,Y=c(23.4,5.4),pop=400,iter=200)
# Aggregation of several trials
for (i in 1:10) {
	output <- rbind(output,dsc(mtcars,reg,Y=c(23.4,5.4),plot=FALSE))
} ; parco(output,"Distance")
# With filtration of min and max y.
output <- dsc(mtcars,reg,Y=c(15,5),ymin=c(14,4),ymax=c(15,6),pop=5000,iter=10000)

Antoine-Masse/KefiR documentation built on Feb. 22, 2024, 5:54 a.m.