addit.gof {xpose4}R Documentation

Additional goodness-of-fit plots, for Xpose 4

Description

This is a compound plot consisting of plots of weighted population residuals (WRES) vs population predictions (PRED), absolute individual weighted residuals (|IWRES|) vs independent variable (IDV), WRES vs IDV, and weighted population residuals vs log(IDV), a specific function in Xpose 4. It is a wrapper encapsulating arguments to the wres.vs.pred, iwres.vs.idv and wres.vs.idv functions.

Usage

addit.gof(
  object,
  type = "p",
  title.size = 0.02,
  title.just = c("center", "top"),
  main = "Default",
  force.wres = FALSE,
  ...
)

Arguments

object

An xpose.data object.

type

1-character string giving the type of plot desired. The following values are possible, for details, see 'plot': '"p"' for points, '"l"' for lines, '"o"' for over-plotted points and lines, '"b"', '"c"') for (empty if '"c"') points joined by lines, '"s"' and '"S"' for stair steps and '"h"' for histogram-like vertical lines. Finally, '"n"' does not produce any points or lines.

title.size

Amount, in a range of 0-1, of how much space the title should take up in the plot)

title.just

how the title should be justified

main

The title of the plot. If "Default" then a default title is plotted. Otherwise the value should be a string like "my title" or NULL for no plot title.

force.wres

Plot the WRES even if other residuals are available.

...

Other arguments passed to link{xpose.plot.default}.

Details

Four additional goodness-of-fit plots are presented side by side for comparison.

A wide array of extra options controlling xyplots are available. See xpose.plot.default and xpose.multiple.plot.default for details.

Value

Returns a compound plot comprising plots of weighted population residuals (WRES) vs population predictions (PRED), absolute individual weighted residuals (|IWRES|) vs independent variable (IDV), WRES vs IDV, and weighted population residuals vs log(IDV).

Author(s)

E. Niclas Jonsson, Mats Karlsson, Andrew Hooker & Justin Wilkins

See Also

wres.vs.pred, iwres.vs.idv, wres.vs.idv, xpose.plot.default, xpose.panel.default, xyplot, xpose.prefs-class, xpose.data-class

Other specific functions: absval.cwres.vs.cov.bw(), absval.cwres.vs.pred(), absval.cwres.vs.pred.by.cov(), absval.iwres.cwres.vs.ipred.pred(), absval.iwres.vs.cov.bw(), absval.iwres.vs.idv(), absval.iwres.vs.ipred(), absval.iwres.vs.ipred.by.cov(), absval.iwres.vs.pred(), absval.wres.vs.cov.bw(), absval.wres.vs.idv(), absval.wres.vs.pred(), absval.wres.vs.pred.by.cov(), absval_delta_vs_cov_model_comp, autocorr.cwres(), autocorr.iwres(), autocorr.wres(), basic.gof(), basic.model.comp(), cat.dv.vs.idv.sb(), cat.pc(), cov.splom(), cwres.dist.hist(), cwres.dist.qq(), cwres.vs.cov(), cwres.vs.idv(), cwres.vs.idv.bw(), cwres.vs.pred(), cwres.vs.pred.bw(), cwres.wres.vs.idv(), cwres.wres.vs.pred(), dOFV.vs.cov(), dOFV.vs.id(), dOFV1.vs.dOFV2(), data.checkout(), dv.preds.vs.idv(), dv.vs.idv(), dv.vs.ipred(), dv.vs.ipred.by.cov(), dv.vs.ipred.by.idv(), dv.vs.pred(), dv.vs.pred.by.cov(), dv.vs.pred.by.idv(), dv.vs.pred.ipred(), gof(), ind.plots(), ind.plots.cwres.hist(), ind.plots.cwres.qq(), ipred.vs.idv(), iwres.dist.hist(), iwres.dist.qq(), iwres.vs.idv(), kaplan.plot(), par_cov_hist, par_cov_qq, parm.vs.cov(), parm.vs.parm(), pred.vs.idv(), ranpar.vs.cov(), runsum(), wres.dist.hist(), wres.dist.qq(), wres.vs.idv(), wres.vs.idv.bw(), wres.vs.pred(), wres.vs.pred.bw(), xpose.VPC(), xpose.VPC.both(), xpose.VPC.categorical(), xpose4-package

Examples

## Here we load the example xpose database 
xpdb <- simpraz.xpdb

## A vanilla plot
addit.gof(xpdb)


[Package xpose4 version 4.7.3 Index]