Display and Analyze ROC Curves • pROC

An R package to display and analyze ROC curves.

For more information, see:

Xavier Robin, Natacha Turck, Alexandre Hainard, et al. (2011) “pROC: an open-source package for R and S+ to analyze and compare ROC curves”. BMC Bioinformatics, 7, 77. DOI: 10.1186/1471-2105-12-77
The official web page
The CRAN page
My blog
The FAQ

Stable

The latest stable version is best installed from the CRAN:

install.packages("pROC")

Getting started

If you don’t want to read the manual first, try the following:

Loading

library(pROC)
data(aSAH)

Basic ROC / AUC analysis

roc(aSAH$outcome, aSAH$s100b)
roc(outcome ~ s100b, aSAH)

Smoothing

roc(outcome ~ s100b, aSAH, smooth=TRUE)

more options, CI and plotting

roc1 <- roc(aSAH$outcome,
            aSAH$s100b, percent=TRUE,
            # arguments for auc
            partial.auc=c(100, 90), partial.auc.correct=TRUE,
            partial.auc.focus="sens",
            # arguments for ci
            ci=TRUE, boot.n=100, ci.alpha=0.9, stratified=FALSE,
            # arguments for plot
            plot=TRUE, auc.polygon=TRUE, max.auc.polygon=TRUE, grid=TRUE,
            print.auc=TRUE, show.thres=TRUE)

    # Add to an existing plot. Beware of 'percent' specification!
    roc2 <- roc(aSAH$outcome, aSAH$wfns,
            plot=TRUE, add=TRUE, percent=roc1$percent)

Coordinates of the curve

coords(roc1, "best", ret=c("threshold", "specificity", "1-npv"))
coords(roc2, "local maximas", ret=c("threshold", "sens", "spec", "ppv", "npv"))

Confidence intervals

# Of the AUC
ci(roc2)

# Of the curve
sens.ci <- ci.se(roc1, specificities=seq(0, 100, 5))
plot(sens.ci, type="shape", col="lightblue")
plot(sens.ci, type="bars")

# need to re-add roc2 over the shape
plot(roc2, add=TRUE)

# CI of thresholds
plot(ci.thresholds(roc2))

Comparisons

    # Test on the whole AUC
    roc.test(roc1, roc2, reuse.auc=FALSE)

    # Test on a portion of the whole AUC
    roc.test(roc1, roc2, reuse.auc=FALSE, partial.auc=c(100, 90),
             partial.auc.focus="se", partial.auc.correct=TRUE)

    # With modified bootstrap parameters
    roc.test(roc1, roc2, reuse.auc=FALSE, partial.auc=c(100, 90),
             partial.auc.correct=TRUE, boot.n=1000, boot.stratified=FALSE)

Sample size

    # Two ROC curves
    power.roc.test(roc1, roc2, reuse.auc=FALSE)
    power.roc.test(roc1, roc2, power=0.9, reuse.auc=FALSE)

    # One ROC curve
    power.roc.test(auc=0.8, ncases=41, ncontrols=72)
    power.roc.test(auc=0.8, power=0.9)
    power.roc.test(auc=0.8, ncases=41, ncontrols=72, sig.level=0.01)
    power.roc.test(ncases=41, ncontrols=72, power=0.9)

Getting Help

Type ?pROC on the R command line
Make sure you’ve read the FAQ
View the FAQ vignette: vignette("FAQ", package = "pROC") or browseVignettes("pROC")
Search for questions tagged with pROC-R-package on Stack Overflow

If you still can’t find an answer, you can:

Development

Building vignettes

To build the vignettes locally:

devtools::build_vignettes()
# Or using tools:
tools::buildVignettes(dir = ".", tangle = FALSE)

The HTML output will be in inst/doc/ after building. After installing the package, view it with:

browseVignettes("pROC")
# Or directly:
vignette("FAQ", package = "pROC")

Installing the development version

Download the source code from git, unzip it if necessary, and then type R CMD INSTALL pROC. Alternatively, you can use the devtools package by Hadley Wickham to automate the process (make sure you follow the full instructions to get started):

if (! requireNamespace("devtools")) install.packages("devtools")
devtools::install_github("xrobin/pROC@develop")

Check

To run all automated tests and R checks, including slow tests:

cd .. # Run from parent directory
VERSION=$(grep Version pROC/DESCRIPTION | sed "s/.\+ //")
R CMD build pROC
RUN_SLOW_TESTS=true R CMD check pROC_$VERSION.tar.gz

Or from an R command prompt with devtools:

devtools::check()

Tests

To run automated tests only from an R command prompt:

run_slow_tests <- TRUE  # Optional, include slow tests
devtools::test()

vdiffr

The vdiffr package is used for visual tests of plots.

To run all the test cases (incl. slow ones) from the command line:

run_slow_tests <- TRUE
devtools::test() # Must run the new tests
testthat::snapshot_review()

To run the checks upon R CMD check, set environment variable NOT_CRAN=1:

NOT_CRAN=1 RUN_SLOW_TESTS=true R CMD check pROC_$VERSION.tar.gz

Release steps

Update Version and Date in DESCRIPTION
Update version and date in NEWS
Get new version to release: VERSION=$(grep Version pROC/DESCRIPTION | sed "s/.\+ //") && echo $VERSION
Build & check package: R CMD build pROC && R CMD check --as-cran pROC_$VERSION.tar.gz
Check with slow tests: NOT_CRAN=1 RUN_SLOW_TESTS=true R CMD check pROC_$VERSION.tar.gz
Check with R-devel: rhub::check_for_cran()
Check reverse dependencies: revdepcheck::revdep_check(num_workers=8, timeout = as.difftime(60, units = "mins"))
Merge into master: git checkout master && git merge develop
Create a tag on master: git tag v$VERSION && git push --tags
Submit to CRAN