StatRocci {plotROC} | R Documentation |
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,
...
)
Arguments
mapping |
Set of aesthetic mappings created by |
data |
The data to be displayed in this layer. There are three options: If A A |
geom |
The geometric object to use to display the data, either as a
|
position |
Position adjustment, either as a string naming the adjustment
(e.g. |
show.legend |
logical. Should this layer be included in the legends?
|
inherit.aes |
If |
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 |
Format
An object of class StatRocci
(inherits from Stat
, ggproto
, gg
) of length 6.
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)))