plotROC (version 2.3.0)

StatRocci: Calculate confidence regions for the empirical ROC curve

Description

Confidence intervals for TPF and FPF are calculated using the exact method of Clopper and Pearson (1934) each at the level 1 - sqrt(1 - alpha). Based on result 2.4 from Pepe (2003), the cross-product of these intervals yields a 1 - alpha

Usage

StatRocci
stat_rocci(
  mapping = NULL,
  data = NULL,
  geom = "rocci",
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  ci.at = NULL,
  sig.level = 0.05,
  na.rm = TRUE,
  ...
)

Format

An object of class StatRocci (inherits from Stat, ggproto, gg) of length 5.

Arguments

mapping

Set of aesthetic mappings created by aes() or aes_(). If specified and inherit.aes = TRUE (the default), it is combined with the default mapping at the top level of the plot. You must supply mapping if there is no plot mapping.

data

The data to be displayed in this layer. There are three options:

If NULL, the default, the data is inherited from the plot data as specified in the call to ggplot().

A data.frame, or other object, will override the plot data. All objects will be fortified to produce a data frame. See fortify() for which variables will be created.

A function will be called with a single argument, the plot data. The return value must be a data.frame, and will be used as the layer data. A function can be created from a formula (e.g. ~ head(.x, 10)).

geom

The geometric object to use display the data

position

Position adjustment, either as a string, or the result of a call to a position adjustment function.

show.legend

logical. Should this layer be included in the legends? NA, the default, includes if any aesthetics are mapped. FALSE never includes, and TRUE always includes. It can also be a named logical vector to finely select the aesthetics to display.

inherit.aes

If FALSE, overrides the default aesthetics, rather than combining with them. This is most useful for helper functions that define both data and aesthetics and shouldn't inherit behaviour from the default plot specification, e.g. borders().

ci.at

Vector of cutoffs at which to display confidence regions. If NULL, will automatically choose 3 evenly spaced points to display the regions

sig.level

Significance level for the confidence regions

na.rm

Remove missing observations

...

Other arguments passed on to layer(). These are often aesthetics, used to set an aesthetic to a fixed value, like colour = "red" or size = 3. They may also be parameters to the paired geom/stat.

Aesthetics

stat_rocci understands the following aesthetics (required aesthetics are in bold):

• m The continuous biomarker/predictor

• d The binary outcome, if not coded as 0/1, the smallest level in sort order is assumed to be 0, with a warning

• alpha

• color

• fill

• linetype

• size

Computed variables

FPF

estimate of false positive fraction

TPF

estimate of true positive fraction

cutoffs

values of m at which estimates are calculated

FPFL

lower bound of confidence region for FPF

FPFU

upper bound of confidence region for FPF

TPFL

lower bound of confidence region for TPF

TPFU

upper bound of confidence region for TPF

References

• Clopper, C. J., and Egon S. Pearson. "The use of confidence or fiducial limits illustrated in the case of the binomial." Biometrika (1934): 404-413.

• Pepe, M.S. "The Statistical Evaluation of Medical Tests for Classification and Prediction." Oxford (2003).

Examples

D.ex <- rbinom(50, 1, .5)
rocdata <- data.frame(D = c(D.ex, D.ex), 
                   M = c(rnorm(50, mean = D.ex, sd = .4), rnorm(50, mean = D.ex, sd = 1)), 
                   Z = c(rep("A", 50), rep("B", 50)))

ggplot(rocdata, aes(m = M, d = D)) + geom_roc() + stat_rocci()
ggplot(rocdata, aes(m = M, d = D)) + geom_roc() + 
stat_rocci(ci.at = quantile(rocdata$M, c(.1, .3, .5, .7, .9)))