plotpd {npde} | R Documentation |
Plots 4 graphs to evaluate the shape of the distribution of prediction discrepancies
plotpd(xobs, pd, ypred)
xobs |
the vector of the observed independent variable (X) |
pd |
the vector of normalised prediction distribution errors (returned
by the functions npde or autonpde with the option output=TRUE) |
ypred |
a vector giving the mean of the predicted distribution for each observation |
Four graphs are produced: \itema quantile-quantile plotplot of the pd versus the corresponding quantiles of a uniform distribution, with the line y=x overlayed. \itema histogram of the pdthe line corresponding to the uniform distribution is also shown \itemtwo scatterplots of the pda plot of the pd versus the independent variable X and a plot of the pd versus the empirical mean of the predicted distribution; for these last two graphs, we plot the lines corresponding to y=0.5, y=0.05 and y=0.95.
None
Emmanuelle Comets <emmanuelle.comets@bichat.inserm.fr>
K. Brendel, E. Comets, C. Laffont, C. Laveille, and F. MentrĂ©. Metrics for external model evaluation with an application to the population pharmacokinetics of gliclazide. Pharmaceutical Research, 23:2036–49, 2006.
#Computing pd data(theopp) data(simtheopp) x<-autonpde(theopp,simtheopp,1,3,4,boolsave=FALSE,calc.pd=TRUE,calc.npde=FALSE) x$pd #Using the npde in object x for the plot plotpd(x$obsdat$xobs,x$pd,x$ypred)