|
1
|
|
# pROC: Tools Receiver operating characteristic (ROC curves) with
|
|
2
|
|
# (partial) area under the curve, confidence intervals and comparison.
|
|
3
|
|
# Copyright (C) 2010-2014 Xavier Robin, Alexandre Hainard, Natacha Turck,
|
|
4
|
|
# Natalia Tiberti, Frédérique Lisacek, Jean-Charles Sanchez
|
|
5
|
|
# and Markus Müller
|
|
6
|
|
#
|
|
7
|
|
# This program is free software: you can redistribute it and/or modify
|
|
8
|
|
# it under the terms of the GNU General Public License as published by
|
|
9
|
|
# the Free Software Foundation, either version 3 of the License, or
|
|
10
|
|
# (at your option) any later version.
|
|
11
|
|
#
|
|
12
|
|
# This program is distributed in the hope that it will be useful,
|
|
13
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
14
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
15
|
|
# GNU General Public License for more details.
|
|
16
|
|
#
|
|
17
|
|
# You should have received a copy of the GNU General Public License
|
|
18
|
|
# along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
19
|
|
|
|
20
|
|
ci.sp <- function(...) {
|
|
21
|
3
|
UseMethod("ci.sp")
|
|
22
|
|
}
|
|
23
|
|
|
|
24
|
|
ci.sp.formula <- function(formula, data, ...) {
|
|
25
|
3
|
data.missing <- missing(data)
|
|
26
|
3
|
roc.data <- roc.utils.extract.formula(formula, data, ...,
|
|
27
|
3
|
data.missing = data.missing,
|
|
28
|
3
|
call = match.call())
|
|
29
|
3
|
if (length(roc.data$predictor.name) > 1) {
|
|
30
|
0
|
stop("Only one predictor supported in 'ci.sp'.")
|
|
31
|
|
}
|
|
32
|
3
|
response <- roc.data$response
|
|
33
|
3
|
predictor <- roc.data$predictors[, 1]
|
|
34
|
3
|
ci.sp(roc(response, predictor, ci=FALSE, ...), ...)
|
|
35
|
|
}
|
|
36
|
|
|
|
37
|
|
ci.sp.default <- function(response, predictor, ...) {
|
|
38
|
3
|
if (methods::is(response, "multiclass.roc") || methods::is(response, "multiclass.auc")) {
|
|
39
|
3
|
stop("'ci.sp' not available for multiclass ROC curves.")
|
|
40
|
|
}
|
|
41
|
3
|
roc <- roc.default(response, predictor, ci = FALSE, ...)
|
|
42
|
3
|
if (methods::is(roc, "smooth.roc")) {
|
|
43
|
0
|
return(ci.sp(smooth.roc = roc, ...))
|
|
44
|
|
}
|
|
45
|
|
else {
|
|
46
|
3
|
return(ci.sp(roc = roc, ...))
|
|
47
|
|
}
|
|
48
|
|
}
|
|
49
|
|
|
|
50
|
|
ci.sp.smooth.roc <- function(smooth.roc,
|
|
51
|
|
sensitivities = seq(0, 1, .1) * ifelse(smooth.roc$percent, 100, 1),
|
|
52
|
|
conf.level = 0.95,
|
|
53
|
|
boot.n = 2000,
|
|
54
|
|
boot.stratified = TRUE,
|
|
55
|
|
progress = getOption("pROCProgress")$name,
|
|
56
|
|
parallel = FALSE,
|
|
57
|
|
...
|
|
58
|
|
) {
|
|
59
|
3
|
if (conf.level > 1 | conf.level < 0)
|
|
60
|
0
|
stop("'conf.level' must be within the interval [0,1].")
|
|
61
|
|
|
|
62
|
3
|
if (roc.utils.is.perfect.curve(smooth.roc)) {
|
|
63
|
0
|
warning("ci.sp() of a ROC curve with AUC == 1 is always a null interval and can be misleading.")
|
|
64
|
|
}
|
|
65
|
|
|
|
66
|
|
# Check if called with density.cases or density.controls
|
|
67
|
3
|
if (is.null(smooth.roc$smoothing.args) || is.numeric(smooth.roc$smoothing.args$density.cases) || is.numeric(smooth.roc$smoothing.args$density.controls))
|
|
68
|
0
|
stop("Cannot compute CI of ROC curves smoothed with numeric density.controls and density.cases.")
|
|
69
|
|
|
|
70
|
|
# Get the non smoothed roc.
|
|
71
|
3
|
roc <- attr(smooth.roc, "roc")
|
|
72
|
3
|
roc$ci <- NULL # remove potential ci in roc to avoid infinite loop with smooth.roc()
|
|
73
|
|
|
|
74
|
|
# prepare the calls
|
|
75
|
3
|
smooth.roc.call <- as.call(c(utils::getS3method("smooth", "roc"), smooth.roc$smoothing.args))
|
|
76
|
|
|
|
77
|
3
|
if(class(progress) != "list")
|
|
78
|
3
|
progress <- roc.utils.get.progress.bar(progress, title="SP confidence interval", label="Bootstrap in progress...", ...)
|
|
79
|
|
|
|
80
|
3
|
if (boot.stratified) {
|
|
81
|
3
|
perfs <- ldply(1:boot.n, stratified.ci.smooth.sp, roc=roc, se=sensitivities, smooth.roc.call=smooth.roc.call, .progress=progress, .parallel=parallel)
|
|
82
|
|
}
|
|
83
|
|
else {
|
|
84
|
3
|
perfs <- ldply(1:boot.n, nonstratified.ci.smooth.sp, roc=roc, se=sensitivities, smooth.roc.call=smooth.roc.call, .progress=progress, .parallel=parallel)
|
|
85
|
|
}
|
|
86
|
|
|
|
87
|
3
|
if (any(is.na(perfs))) {
|
|
88
|
0
|
warning("NA value(s) produced during bootstrap were ignored.")
|
|
89
|
0
|
perfs <- perfs[!apply(perfs, 1, function(x) any(is.na(x))),]
|
|
90
|
|
}
|
|
91
|
|
|
|
92
|
3
|
ci <- t(apply(perfs, 2, quantile, probs=c(0+(1-conf.level)/2, .5, 1-(1-conf.level)/2)))
|
|
93
|
3
|
rownames(ci) <- paste(sensitivities, ifelse(roc$percent, "%", ""), sep="")
|
|
94
|
|
|
|
95
|
3
|
class(ci) <- c("ci.sp", "ci", class(ci))
|
|
96
|
3
|
attr(ci, "conf.level") <- conf.level
|
|
97
|
3
|
attr(ci, "boot.n") <- boot.n
|
|
98
|
3
|
attr(ci, "boot.stratified") <- boot.stratified
|
|
99
|
3
|
attr(ci, "sensitivities") <- sensitivities
|
|
100
|
3
|
attr(ci, "roc") <- smooth.roc
|
|
101
|
3
|
return(ci)
|
|
102
|
|
}
|
|
103
|
|
|
|
104
|
|
ci.sp.roc <- function(roc,
|
|
105
|
|
sensitivities = seq(0, 1, .1) * ifelse(roc$percent, 100, 1),
|
|
106
|
|
conf.level = 0.95,
|
|
107
|
|
boot.n = 2000,
|
|
108
|
|
boot.stratified = TRUE,
|
|
109
|
|
progress = getOption("pROCProgress")$name,
|
|
110
|
|
parallel = FALSE,
|
|
111
|
|
...
|
|
112
|
|
) {
|
|
113
|
3
|
if (conf.level > 1 | conf.level < 0)
|
|
114
|
0
|
stop("'conf.level' must be within the interval [0,1].")
|
|
115
|
|
|
|
116
|
3
|
if (roc.utils.is.perfect.curve(roc)) {
|
|
117
|
0
|
warning("ci.sp() of a ROC curve with AUC == 1 is always a null interval and can be misleading.")
|
|
118
|
|
}
|
|
119
|
|
|
|
120
|
3
|
if(class(progress) != "list")
|
|
121
|
3
|
progress <- roc.utils.get.progress.bar(progress, title="SP confidence interval", label="Bootstrap in progress...", ...)
|
|
122
|
|
|
|
123
|
3
|
if (boot.stratified) {
|
|
124
|
3
|
perfs <- ldply(1:boot.n, stratified.ci.sp, roc=roc, se=sensitivities, .progress=progress, .parallel=parallel)
|
|
125
|
|
}
|
|
126
|
|
else {
|
|
127
|
3
|
perfs <- ldply(1:boot.n, nonstratified.ci.sp, roc=roc, se=sensitivities, .progress=progress, .parallel=parallel)
|
|
128
|
|
}
|
|
129
|
|
|
|
130
|
3
|
if (any(is.na(perfs))) {
|
|
131
|
0
|
warning("NA value(s) produced during bootstrap were ignored.")
|
|
132
|
0
|
perfs <- perfs[!apply(perfs, 1, function(x) any(is.na(x))),]
|
|
133
|
|
}
|
|
134
|
|
|
|
135
|
3
|
ci <- t(apply(perfs, 2, quantile, probs=c(0+(1-conf.level)/2, .5, 1-(1-conf.level)/2)))
|
|
136
|
3
|
rownames(ci) <- paste(sensitivities, ifelse(roc$percent, "%", ""), sep="")
|
|
137
|
|
|
|
138
|
3
|
class(ci) <- c("ci.sp", "ci", class(ci))
|
|
139
|
3
|
attr(ci, "conf.level") <- conf.level
|
|
140
|
3
|
attr(ci, "boot.n") <- boot.n
|
|
141
|
3
|
attr(ci, "boot.stratified") <- boot.stratified
|
|
142
|
3
|
attr(ci, "sensitivities") <- sensitivities
|
|
143
|
3
|
attr(ci, "roc") <- roc
|
|
144
|
3
|
return(ci)
|
|
145
|
|
}
|
xrobin