nm.covplot: Create covariate plot and data

In qPharmetra/qpToolkit: qPharmetra R Tools

Create covariate plot and data

Description

Create ETA vs covariates and parameter estimates vs covariates. Programmed with the intention to get between 95-100% of report Appendices ready plots. If not, the data.frames on basis of which they were created are available as well to enable creating the desired graphs very easily.

Usage

nm.covplot(
  run = "run1",
  path = getOption("nmDir"),
  id.var = "ID",
  iov = FALSE,
  eta.densities = TRUE,
  eta.skip = NULL,
  shrinkage = FALSE,
  covlist = list(cat = c(NULL), con = c(NULL)),
  catcov.sep = 5,
  parameters = NULL,
  aspect = "fill",
  pcx = 0.5,
  uniques = FALSE,
  which.list = 1
)

Arguments

run	run rootname (e.g. run1)
path	directory where run resides
id.var	character string denoting grouping variable ("ID")
iov	inter-occasion variability flag (this argument is ignored currently)
eta.densities	logical to determine of ETA density plots have to be generated
eta.skip	character vector of ETAs that will be skipped
shrinkage	providing shrinkage values. This is currently ignored
covlist	a list with elements cat and con each representing a character vector with categorical and continuous covariates, respectively.
catcov.sep	max number of unique values for a covariate to be assigned 'categorical'
parameters	a character vector with parameter names to plot
aspect	lattice banking aspect for rectangular or square plots. Defaults to 'fill'
pcx	dot size
uniques	set to TRUE to look at one value per ID only. defaults to F
which.list	which element of the list in case multiple regression methods were used. Defaults to 1.

Value

A list with plots and properly sorted / molten data.frames for customized plots

Examples

args(nm.covplot)
nmcov = nm.covplot(path = getOption("qpExampleDir")
, run = "example2", parameters = c("CL","V1","Q","V2")
, covlist = list(con = "AGE", cat = "GNDR"))

#contents of the list
names(nmcov)
#[1] "covdata"        "eta.splom"      "eta.dens"
# "conData.eta"    "catData.eta"    "conData.par"    "catData.par"
#[8] "etaContVarPlot" "etaCatVarPlot"  "parContVarPlot" "parCatVarPlot"

# let's look at the plots
nmcov$eta.dens()
# red line is the distribution predicted obo OMEGA estimate
# the grey polygon is the actual ETA distribution

## elegant pairs plot
nmcov$eta.splom()

## ETAs vs. Continuous Covariates
nmcov$etaContVarPlot()

## ETAs vs. Categorical Covariates
nmcov$etaCatVarPlot()
nmcov$etaCatVarPlot(F)

## parameters vs. Continuous Covariates
nmcov$parContVarPlot()

## parameters vs. Categorical Covariates
nmcov$parCatVarPlot()

## data frames required to replot the above are right there for convenience
head(nmcov$covdata) ## merged get.xpose.table and nm.params.table output!
head(nmcov$conData.eta)# essential: conVariable conValue variable value
head(nmcov$catData.eta)# essential: catVariable catValue variable value
head(nmcov$conData.par)# essential: conVariable conValue variable value
head(nmcov$catData.par)# essential: catVariable catValue variable value

#for continuous plots get kickstarted with:
# xyplot(value ~ conValue | casefold(variable, upper = TRUE) * conVariable
#  , data = conData, panel = panel.xyplot)
#for categorical plots:
# xyplot(value ~ as.factor(paste(catValue)) | variable * catVariable
#  , data = catData, panel = panel.bwplot, horizontal = FALSE)

qPharmetra/qpToolkit documentation built on May 24, 2023, 8:52 a.m.