Comparative probability plots for parametric survival models

DESCRIPTION:

Comparative probability plots for parametric survival models based upon a censored regression model.

USAGE:

probplot6.censorReg(object, plot.at=NULL, method=c("KM", "separate", "factor", 
     "regression", "one", "none"), distribution=c("weibull","lognormal",  
     "loglogistic"), mle.interval=F, conf.level=0.95, xlab=NULL,  
     ylab="Probability", main.title=NULL, xlim=NULL, ylim=NULL,  
     add.legend=F) 

REQUIRED ARGUMENTS:

object
An object of class "censorReg".

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
in addition to probabilities for the model, probabilities for one further 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) 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 x "one" censor(y) ~ 1 censorReg model with no covariates "none" no points are plotted The default is method="KM".
distribution
the distributions for which comparative plots are desired. This may be one or more of "weibull", "lognormal", "loglogistic". Plots are made for both the logged and the unlogged versions of the distribution.
mle.interval
logical value indicating whether add confidence intervals for the maximum likelihood estimate (mle).
conf.level
the confidence level for confidence intervals.
xlab
character string specifying the x-axis label.
ylab
character string specifying the y-axis label. The default is "Probability".
xlim
numeric vector of length two specifying the x-axis limits on the scale of response(object).
ylim
numeric vector of length two specifying the y-axis limits on the probability scale.
add.legend
logical value indicating whether or not to add a legend to each plot using the locator function. This requires clicking in each graph for placing the upper left corner of the legend.

SIDE EFFECTS::

A probability plot is produced on the current graphics device. This is a multiple figure plot with up to six figures, one for each distribution.

SEE ALSO:

, ,

EXAMPLES:

probplot6.censorReg(censorReg(censor(days,event)~voltage, data=capacitor2, weights =  
    weights))