1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
|
\encoding{UTF-8}
\name{ci.thresholds}
\alias{ci.thresholds}
\alias{ci.thresholds.default}
\alias{ci.thresholds.formula}
\alias{ci.thresholds.roc}
\alias{ci.thresholds.smooth.roc}
\title{
Compute the confidence interval of thresholds
}
\description{
This function computes the confidence interval (CI) of the sensitivity
and specificity of the thresholds given in argument.
By default, the 95\% CI are computed with 2000 stratified bootstrap replicates.
}
\usage{
# ci.thresholds(...)
\S3method{ci.thresholds}{roc}(roc, conf.level=0.95, boot.n=2000,
boot.stratified=TRUE, thresholds = "local maximas",
progress=getOption("pROCProgress")$name, parallel=FALSE, ...)
\S3method{ci.thresholds}{formula}(formula, data, ...)
\S3method{ci.thresholds}{smooth.roc}(smooth.roc, ...)
\S3method{ci.thresholds}{default}(response, predictor, ...)
}
\arguments{
\item{roc}{a \dQuote{roc} object from the \code{\link{roc}} function.}
\item{smooth.roc}{not available for \link[=smooth.roc]{smoothed} ROC
curves, available only to catch the error and provide a clear error
message.
}
\item{response, predictor}{arguments for the \code{\link{roc}} function.}
\item{formula, data}{a formula (and possibly a data object) of type
response~predictor for the \code{\link{roc}} function.
}
\item{conf.level}{the width of the confidence interval as [0,1], never
in percent. Default: 0.95, resulting in a 95\% CI.
}
\item{boot.n}{the number of bootstrap replicates. Default: 2000.}
\item{boot.stratified}{should the bootstrap be stratified (default, same number
of cases/controls in each replicate than in the original sample) or
not.
}
\item{thresholds}{on which thresholds to evaluate the CI. Either the
numeric values of the thresholds, a logical vector (as index of
\code{roc$thresholds}) or a character \dQuote{all}, \dQuote{local
maximas} or \dQuote{best} that will be used to determine the threshold(s)
on the supplied curve with \code{\link{coords}} (not on the resampled curves).
}
\item{progress}{the name of progress bar to display. Typically
\dQuote{none}, \dQuote{win}, \dQuote{tk} or \dQuote{text} (see the
\code{name} argument to \code{\link[plyr]{create_progress_bar}} for
more information), but a list as returned by \code{\link[plyr]{create_progress_bar}}
is also accepted. See also the \dQuote{Progress bars} section of
\link[=pROC-package]{this package's documentation}.
}
\item{parallel}{if TRUE, the bootstrap is processed in parallel, using
parallel backend provided by plyr (foreach).
}
\item{\dots}{further arguments passed to or from other methods,
especially arguments for \code{\link{roc}} and \code{ci.thresholds.roc}
when calling \code{ci.thresholds.default} or \code{ci.thresholds.formula}.
Arguments for \code{\link{txtProgressBar}} (only
\code{char} and \code{style}) if applicable.
Arguments \code{best.method} and \code{best.weights} to \code{\link{coords}}.
}
}
\details{
\code{ci.thresholds.formula} and \code{ci.thresholds.default} are convenience methods
that build the ROC curve (with the \code{\link{roc}} function) before
calling \code{ci.thresholds.roc}. You can pass them arguments for both
\code{\link{roc}} and \code{ci.thresholds.roc}. Simply use \code{ci.thresholds}
that will dispatch to the correct method.
This function creates \code{boot.n} bootstrap replicate of the ROC
curve, and evaluates the sensitivity and specificity at thresholds
given by the \code{thresholds} argument. Then it computes the
confidence interval as the percentiles given by \code{conf.level}.
A threshold given as a \code{logical} vector or \code{character} is converted to the corresponding numeric vector once
\emph{using the supplied ROC curve}, and not at each bootstrap iteration. See \code{\link{ci.coords}} for the latter behaviour.
For more details about the bootstrap, see the Bootstrap section in
\link[=pROC-package]{this package's documentation}.
}
\section{Warnings}{
If \code{boot.stratified=FALSE} and the sample has a large imbalance between
cases and controls, it could happen that one or more of the replicates
contains no case or control observation, producing a \code{NA} area.
The warning \dQuote{NA value(s) produced during bootstrap were ignored.}
will be issued and the observation will be ignored. If you have a large
imbalance in your sample, it could be safer to keep
\code{boot.stratified=TRUE}.
}
\value{
A list of length 2 and class \dQuote{ci.thresholds}, \dQuote{ci} and \dQuote{list} (in this order), with the confidence
intervals of the CI and the following items:
\item{specificity}{a matrix of CI for the specificity. Row (names) are the
thresholds, the first column the lower bound, the 2nd column the
median and the 3rd column the upper bound.
}
\item{sensitivity}{same than specificity.}
Additionally, the list has the following attributes:
\item{conf.level}{the width of the CI, in fraction.}
\item{boot.n}{the number of bootstrap replicates.}
\item{boot.stratified}{whether or not the bootstrapping was stratified.}
\item{thresholds}{the thresholds, as given in argument.}
\item{roc}{the object of class \dQuote{\link{roc}} that was used to
compute the CI.
}
}
\references{
James Carpenter and John Bithell (2000) ``Bootstrap condence intervals:
when, which, what? A practical guide for medical statisticians''.
\emph{Statistics in Medicine} \bold{19}, 1141--1164.
DOI: \doi{10.1002/(SICI)1097-0258(20000515)19:9<1141::AID-SIM479>3.0.CO;2-F}.
Tom Fawcett (2006) ``An introduction to ROC analysis''. \emph{Pattern
Recognition Letters} \bold{27}, 861--874. DOI:
\doi{10.1016/j.patrec.2005.10.010}.
Xavier Robin, Natacha Turck, Alexandre Hainard, \emph{et al.}
(2011) ``pROC: an open-source package for R and S+ to analyze and
compare ROC curves''. \emph{BMC Bioinformatics}, \bold{7}, 77.
DOI: \doi{10.1186/1471-2105-12-77}.
Hadley Wickham (2011) ``The Split-Apply-Combine Strategy for Data Analysis''. \emph{Journal of Statistical Software}, \bold{40}, 1--29.
URL: \doi{10.18637/jss.v040.i01}.
}
\seealso{
\code{\link{roc}},
\code{\link{ci}}
}
\examples{
data(aSAH)
# Create a ROC curve:
data(aSAH)
roc1 <- roc(aSAH$outcome, aSAH$s100b)
## Basic example ##
# Compute CI of all local maxima thresholds
\dontrun{
ci.thresholds(roc1)}\dontshow{ci.thresholds(roc1, boot.n = 10)}
## More options ##
# Customized bootstrap and thresholds:
\dontrun{
ci.thresholds(roc1,
thresholds=c(0.5, 1, 2),
boot.n=10000, conf.level=0.9, stratified=FALSE)}\dontshow{
ci.thresholds(roc1,
thresholds=c(0.5, 1, 2),
boot.n=10, conf.level=0.9, stratified=FALSE)}
## Plotting the CI ##
\dontrun{
ci1 <- ci.thresholds(roc1)}\dontshow{
ci1 <- ci.thresholds(roc1, boot.n = 10)}
plot(roc1)
plot(ci1)
}
\keyword{univar}
\keyword{nonparametric}
\keyword{utilities}
\keyword{roc}
|
