plotMultiCoxph: Plot `multi.coxph` object

Description

Plot multi.coxph object

Usage

## S3 method for class 'multi.coxph':
plot(x, ...,
    type = c("p", "w", "s"))

Arguments

An object of class multi.coxph

...

Not implemented

type

Type of plot

Value

A graph (base graphics).

Details

One of three types of graph is possible. If type="p" then points representing the information criterion (IC) for each model are plotted. A line is also drawn 2 units above the minimum IC. Models below this are typically worth considering. If it is not visible on the plot, then important models have have been overlooked, suggesting a larger value for confSetSize may be appropriate. If type="w" then the weights (relative evidence weights) of the models are plotted. These can be interpreted as the probability that each model is the best in the set. A red vertical line is shown where the cumulated evidence weight reaches 95 If type="s" then the sum of the relative evidence weights for each term/coefficient is plotted. The sum is taken across all models in which the term appears.

Examples

Run this code

set.seed(1)
dt1 <- genSurvDt(b=2, c=5, f=0, model=FALSE)
m1 <- multi(coxph(Surv(t1, e) ~ ., data=dt1), crit="bic")
plot(m1, type="w")