Create A Probability Plot for a Parametric Survival Model

DESCRIPTION:

Create a probability plot for a parametric survival model with one or fewer covariates or stratification variables.

USAGE:

probplot.censorReg(object, plot.at=unique(group.variable),  
   method=c("KM", "separate", "factor", "regression", "one", "none"),  
   xlab=paste("Time to Failure - ", object$threshold),  
   ylab=paste(distribution2, "Probability"), main.title=NULL, sub=NULL, 
   conf.level=0.95, xlim=NULL, ylim=NULL, mle=T, mle.interval=F, upper=T,  
   ower=T, pch.data=1:lobj, lty.mle=1:lobj, lty.interval=1:lobj,  
   col.data=1:lobj, col.mle=1:lobj, col.interval=1:lobj, add.legend=F,  
   data=NULL) 

REQUIRED ARGUMENTS:

object
an object of class "censorReg" or "censorRegList" containing one or fewer covariates or stratification variables.

OPTIONAL ARGUMENTS:

plot.at
the values of the covariate for which a probability distribution is to be plotted. Ignored if there is no covariate in the model in object.
method
character string specifying method for computing the quantiles of the plotted points on the graph. In addition to probabilities for the model in object, probabilities for one other model can be plotted as points. Let the covariate be x, and let the response be censor(y). Then the possible methods for these alternative models are: "KM" censor(y) ~ strata(x) Turnbull's generalization of Kaplan-Meier estimates "separate" censor(y) ~ strata(x) censorReg estimates with separate location and scale for each covariate value. "factor" censor(y) ~ factor(x) censorReg estimates with separate location and the same scale for each covariate value "regression" censor(y) ~ x censorReg model with location predicted by regression on the covariate "one" censor(y) ~ 1 censorReg model with no covariates "none" The comparison points are not plotted. The default is method="KM".
xlab
character string specifying the x-axis label.
ylab
character string specifying the y-axis label.
main.title
character string specifying the main title. If NULL, an appropriate main title is generated. Use main.title="" for no main title.
sub
character string specifying a sub title displayed below the x-axis label. If not specified, and date.it = T this will default to the date the object was created.
conf.level
the confidence level for plotting confidence intervals of the estimated model probabilities.
xlim
numeric vector specifying the x-axis limits on the scale of the response.
ylim
numeric vector specifying the y-axis limits on the probability scale.
mle
logical value indicating whether or not to add the maximum likelihood estimate(s) of the estimated distribution(s).
mle.interval
logical value indicating whether or not to add confidence intervals around the maximum likelihood estimate(s).
upper
logical value indicating whether or not to add the upper confidence limit for the maximum likelihood estimate(s).
lower
logical value indicating whether or not to add the lower confidence limit for the maximum likelihood estimate(s).
pch.data
vector of indices for plot symbols. The i-th element in pch.data contains the plot symbol index (see pch in function par) for the i-th probability distribution. The plot symbol is used to display the alternative model probabilities.
lty.mle
vector of line types for the maximum likelihood estimate of the failure probability. The i-th element in lty.mle contains the line type (see lty in function par) for the i-th probability distribution.
lty.interval
vector of line types for the maximum likelihood confidence interval estimates. The i-th element in lty.interval contains the line type (see lty in function par) for the i-th set of confidence intervals.
col.data
vector of color indices for plot symbols. The i-th element in col.data contains the color index (see col in function par) for the i-th probability distribution. The color is used to display the alternative model probabilities.
col.mle
vector of color types for the maximum likelihood estimate of the failure probability. The i-th element in col.mle contains the color type (see col in function par) for the i-th probability distribution.
col.interval
vector of color types for the maximum likelihood confidence intervals. The i-th element in col.interval contains the color type (see col in function par) for the i-th set of confidence intervals.
add.legend
if T, a legend is added to the plot (using the locator() function). The legend gives the quantiles associated with each line and color type.
legend.loc
location for placing the upper left corner of the legend. If missing, you must click in the graph to locate the legend. May also specify a vector of values for locating the legend or the string "auto" for automatic placement.

VALUE:

the values plotted for the alternative distribution(s).

DETAILS:

This function is a method for the generic function plot for class "censorReg" . It can be invoked by calling probplot for an object of the appropriate class.

SIDE EFFECTS::

A probability plot of the fit in object is produced on the current graphics device.

SEE ALSO:

censorReg, , qqplot.censorReg.setup, qqplot.censorReg.plot.quantiles

EXAMPLES:

a <- censorReg(censor(days,event) ~ voltage, data=capacitor2, weights = weights) 
probplot.censorReg(a) 
probplot.censorReg(a, plot.at=c(26,32), method="separate")