Skip to contents

Plot Random Hazard Forest hazard / cumulative-hazard curves

Source: R/plot.gg_rhf.R
plot.gg_rhf.Rd

Draws case-specific ensemble curves from a gg_rhf() object: hazard (default) or cumulative hazard, one line per case selected by idx.

Usage

# S3 method for class 'gg_rhf'
plot(x, idx = NULL, hazard.only = TRUE, ...)

Arguments

x

A gg_rhf object from gg_rhf().

idx

Integer vector of case ids (matched against the id column) to draw. NULL (default) draws every case.

hazard.only

Logical; TRUE (default) plots the hazard, FALSE plots the cumulative hazard.

...

Not currently used.

Value

A ggplot object.

Details

Cells the forest left undefined are dropped before drawing. From randomForestRHF 2.0.0 the hazard is NA outside each case's observed (start, stop] path, so a hazard curve ends with that case's follow-up, and a case whose hazard is masked throughout is left out of the legend rather than shown as an empty one. From 2.0.3 the cumulative hazard is masked too, though on a different rule: it is NA after each case's final stop, and stays flat through an internal gap. So hazard.only = FALSE also ends each curve with that case's follow-up, but it keeps the grid points a gap would have cost the hazard panel.

Each case gets its own colour and legend key only while there are at most 10 of them. Beyond that the colours stop being distinguishable and the legend would list every id, so the curves are drawn in a single translucent colour with no legend. To pick out individual cases, pass a short idx.

See also

gg_rhf().

Examples

# \donttest{
if (requireNamespace("randomForestRHF", quietly = TRUE)) {
  data(pbc, package = "randomForestSRC")
  d <- randomForestRHF::convert.counting(
    survival::Surv(days, status) ~ ., na.omit(pbc))
  o <- randomForestRHF::rhf("Surv(id, start, stop, event) ~ .", d, ntree = 30)
  plot(gg_rhf(o), idx = c(1, 5, 10))
}

# }