plot.pec: Plotting prediction error curves

Description

Plotting prediction error curves for one or more prediction models.

Usage

## S3 method for class 'pec':
plot(x,
    what,
    models,
    xlim=c(x$start,x$minmaxtime),
    ylim=c(0,0.3),
    xlab="Time",
    ylab,
    axes=TRUE,
    col,
    lty,
    lwd,
    type,
    smooth=FALSE,
    add.refline=FALSE,
    add=FALSE,
    legend=ifelse(add,FALSE,TRUE),
    special=FALSE,
     ...)

Arguments

Object of class pec obtained with function pec.

what

The name of the entry in x. Defauls to PredErr Other choices are AppErr, BootCvErr, Boot632, Boot632plus.

models

Specifies models in x$models for which the prediction error curves are drawn. Defaults to all models.

xlim

Plotting range on the x-axis.

ylim

Plotting range on the y-axis.

xlab

Label given to the x-axis.

ylab

Label given to the y-axis.

axes

Logical. If FALSE no axes are drawn.

col

Vector of colors given to the curves of models in the order determined by models.

lty

Vector of lty's given to the curves of models in the order determined by models.

lwd

Vector of lwd's given to the curves of models in the order determined by models.

type

Plotting type: either "l" or "s", see lines.

smooth

Logical. If TRUE the plotting values are smoothed with the function smooth kind="3R".

add.refline

Logical. If TRUE a dotted horizontal line is drawn as a symbol for the naive rule that predicts probability .5 at all cutpoints (i.e. time points in survival analysis).

add

Logical. If TRUE only lines are added to an existing device

legend

if TRUE a legend is plotted by calling the function legend. Optional arguments of the function legend can be given in the form legend.x=val where x is the name of the argument and val the desired value. See also Deta

special

Logical. If TRUE the bootstrap curves of models are plotted together with predErr of models by invoking the function Special. Optional arguments of the function Special

...

Extra arguments that are passed to plot.

Value

The (invisible) object.

Details

From version 2.0.1 on the arguments legend.text, legend.args, lines.type, lwd.lines, specials are obsolete and only available for backward compatibility. Instead arguments for the invoked functions legend, axis, Special are simply specified as legend.lty=2. The specification is not case sensitive, thus Legend.lty=2 or LEGEND.lty=2 will have the same effect. The function axis is called twice, and arguments of the form axis1.labels, axis1.at are used for the time axis whereas axis2.pos, axis1.labels, etc. are used for the y-axis.

These arguments are processed via ... of plot.pec and inside by using the function resolveSmartArgs. Documentation of these arguments can be found in the help pages of the corresponding functions.

Examples

Run this code

# simulate an artificial data frame
# with survival response and two predictors

set.seed(280180)
N <- 300
X1 <- rnorm(N,mean=10,sd=5)
X2 <- rbinom(N,1,.4)
linPred <- .00001+0.2*X1+2.3*X2
T <- sapply(linPred,function(lp){rexp(n=1,exp(-lp))})
C <- rexp(n=300,exp(-mean(linPred)))
dat <- data.frame(time=pmin(T,C),status=as.numeric(T<=C),X1=X1,X2=X2)

# fit some candidate Cox models and compute the Kaplan-Meier estimate 

Models <- list("Kaplan.Meier"=survfit(Surv(time,status)~1,data=dat),
               "Cox.X1"=coxph(Surv(time,status)~X1,data=dat),
               "Cox.X2"=coxph(Surv(time,status)~X2,data=dat),
               "Cox.X1.X2"=coxph(Surv(time,status)~X1+X2,data=dat))


# compute the .632+ estimate of the generalization error 
set.seed(17100)
PredError.632plus <- pec(object=Models,
                         formula=Surv(time,status)~X1+X2,
                         data=dat,
                         exact=TRUE,
                         cens.model="marginal",
                         splitMethod="boot632plus",
                         B=10,
                         keep.matrix=TRUE,
                         verbose=TRUE)

# plot the .632+ estimates of the generalization error 
plot(PredError.632plus,xlim=c(0,30))

# plot the bootstrapped curves, .632+ estimates of the generalization error
# and Apparent error for the Cox model 'Cox.X1' with the 'Cox.X2' model
# as benchmark
plot(PredError.632plus,xlim=c(0,30), models="Cox.X1", special=TRUE, special.bench="Cox.X2", special.benchcol=2, special.addprederr="AppErr")

Run the code above in your browser using DataLab

Get 50% off unlimited learning