Package 'plotROC' reference manual

Title:	Generate Useful ROC Curve Charts for Print and Interactive Use
Description:	Most ROC curve plots obscure the cutoff values and inhibit interpretation and comparison of multiple curves. This attempts to address those shortcomings by providing plotting and interactive tools. Functions are provided to generate an interactive ROC curve plot for web use, and print versions. A Shiny application implementing the functions is also included.
Authors:	Michael C. Sachs [aut, cre], Robert W. Corty [ctb], Luis Crouch [ctb] (Modification to calc_auc)
Maintainer:	Michael C. Sachs <[email protected]>
License:	MIT + file LICENSE
Version:	2.3.2
Built:	2024-11-14 23:27:00 UTC
Source:	https://github.com/sachsmc/plotROC

Calculate the Area under the ROC curve

Description

Given a ggplot object with a GeomRoc layer, computes the area under the ROC curve for each group

Usage

calc_auc(ggroc)
calc_auc(ggroc)

Arguments

ggroc

A ggplot object that contains a GeomRoc layer

Value

A data frame with the estimated AUCs for each layer, panel and group

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)))

ggroc <- ggplot(rocdata, aes(m = M, d = D)) + geom_roc()
calc_auc(ggroc)
ggroc2 <- ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc()
calc_auc(ggroc2)
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)))

ggroc <- ggplot(rocdata, aes(m = M, d = D)) + geom_roc()
calc_auc(ggroc)
ggroc2 <- ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc()
calc_auc(ggroc2)

Calculate the Empirical ROC curves for multiple biomarkers

Description

Deprecated, use geom_roc instead

Usage

calculate_multi_roc(data, M_string, D_string)
calculate_multi_roc(data, M_string, D_string)

Arguments

`data`	data frame containing at least 1 marker and the common class labels, coded as 0 and 1
`M_string`	vector of marker column names
`D_string`	class label column name

Value

List of data frames containing cutoffs, and estimated true and false positive fractions

Calculate the Empirical ROC curve

Description

Deprecated, use geom_roc instead

Usage

calculate_roc(M, D, ci = FALSE, alpha = 0.05)
calculate_roc(M, D, ci = FALSE, alpha = 0.05)

Arguments

`M`	continuous marker values or predictions of class labels
`D`	class labels, must be coded as 0 and 1. If not numeric with 0/1, then plotROC assumes the first level in sort order is healthy status, with a warning.
`ci`	Logical, if true, will calculate exact joint confidence regions for the TPF and FPF
`alpha`	Confidence level, ignored if `ci = FALSE`

Details

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

Value

A dataframe containing cutoffs, estimated true and false positive fractions, and confidence intervals if ci = TRUE.

Add direct labels to a ROC plot

Description

Add direct labels to a ROC plot

Usage

direct_label(
  ggroc_p,
  labels = NULL,
  label.angle = 45,
  nudge_x = 0,
  nudge_y = 0,
  size = 6,
  ...
)
direct_label(
  ggroc_p,
  labels = NULL,
  label.angle = 45,
  nudge_x = 0,
  nudge_y = 0,
  size = 6,
  ...
)

Arguments

`ggroc_p`	A ggplot object that contains a geom_roc layer
`labels`	vector of labels to add directly to the plot next to the curves. If multiple curves, must be in the same order as the grouping factor. If NULL, attempts to determine labels from the ggroc_p object
`label.angle`	angle of adjustment for the direct labels
`nudge_x`, `nudge_y`	Horizontal and vertical adjustment to nudge labels by. These can be scalars or vectors the same length as the number of labels
`size`	Size of labels
`...`	Other arguments passed to annotate

Generate svg code for an ROC curve object

Description

Takes a ggplot object that contains a GeomRoc layer and returns a string that contains html suitable for creating a standalone interactive ROC curve plot.

Usage

export_interactive_roc(
  ggroc_p,
  add.cis = TRUE,
  hide.points = FALSE,
  prefix = "a",
  width = 6,
  height = 6,
  omit.js = FALSE,
  style = style_roc(theme = theme_grey()),
  ...
)
export_interactive_roc(
  ggroc_p,
  add.cis = TRUE,
  hide.points = FALSE,
  prefix = "a",
  width = 6,
  height = 6,
  omit.js = FALSE,
  style = style_roc(theme = theme_grey()),
  ...
)

Arguments

`ggroc_p`	A ggplot object with a GeomRoc layer and optionally a GeomRocci layer as returned by geom_roc and/or geom_rocci. It can be modified with annotations, themes, etc.
`add.cis`	Logical, if true, removes the current confidence interval layer (if present) and replaces it with a denser layer of confidence regions
`hide.points`	Logical, if true, hides points layer so that points with cutoff values are only visible when hovering. Recommended for plots containing more than 3 curves.
`prefix`	A string to assign to the objects within the svg. Enables unique identification by the javascript code
`width`, `height`	Width and height in inches of plot
`omit.js`	Logical. If true, omit inclusion of javascript source in output. Useful for documents with multiple interactive plots
`style`	A call to the function style_roc
`...`	Other arguments passed to geom_rocci when `add.cis = TRUE`

Details

If you intend to include more than one of these objects in a single page, use a different prefix string for each one. To use this function in knitr, use the chunk options fig.keep='none' and results = 'asis', then cat() the resulting string to the output. See the vignette for examples. Older browsers (< IE7) are not supported.

Value

A character object containing the html necessary to plot the ROC curve in a web browser

Confidence regions for the ROC curve

Description

Display rectangular confidence regions for the empirical ROC curve.

Usage

geom_rocci(
  mapping = NULL,
  data = NULL,
  stat = "rocci",
  ci.at = NULL,
  sig.level = 0.05,
  na.rm = TRUE,
  alpha.box = 0.3,
  labels = TRUE,
  labelsize = 3.88,
  labelround = 1,
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  ...
)

GeomRocci
geom_rocci(
  mapping = NULL,
  data = NULL,
  stat = "rocci",
  ci.at = NULL,
  sig.level = 0.05,
  na.rm = TRUE,
  alpha.box = 0.3,
  labels = TRUE,
  labelsize = 3.88,
  labelround = 1,
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  ...
)

GeomRocci

Arguments

`mapping`	Set of aesthetic mappings created by `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)`).
`stat`	Use to override the default connection between `geom_rocci` and `stat_rocci`.
`ci.at`	Vector of values in the range of the biomarker where confidence regions will be displayed
`sig.level`	Significance level for the confidence regions
`na.rm`	If `FALSE`, the default, missing values are removed with a warning. If `TRUE`, missing values are silently removed.
`alpha.box`	Alpha level for the confidence regions
`labels`	If TRUE, adds text labels for the cutoffs where the confidence regions are displayed
`labelsize`	Size of cutoff text labels
`labelround`	Integer, number of significant digits to round cutoff labels
`position`	Position adjustment, either as a string naming the adjustment (e.g. `"jitter"` to use `position_jitter`), or the result of a call to a position adjustment function. Use the latter if you need to change the settings of the adjustment.
`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()`.
`...`	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.

Format

An object of class GeomRocci (inherits from Geom, ggproto, gg) of length 6.

Aesthetics

geom_rocci understands the following aesthetics (required aesthetics are in bold). stat_rocci automatically maps the estimates to the required aesthetics:

x The FPF estimate
y The TPF estimate
xmin Lower confidence limit for the FPF
xmax Upper confidence limit for the FPF
ymin Lower confidence limit for the TPF
ymax Upper confidence limit for the TPF
alpha
color
fill
linetype
size

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() + geom_rocci()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(sig.level = .01)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(n.cuts = 0) + 
geom_rocci(ci.at = quantile(rocdata$M, c(.1, .25, .5, .75, .9)))
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(linetype = 1)
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() + geom_rocci()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(sig.level = .01)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(n.cuts = 0) + 
geom_rocci(ci.at = quantile(rocdata$M, c(.1, .25, .5, .75, .9)))
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(linetype = 1)

Empirical Receiver Operating Characteristic Curve

Description

Display the empirical ROC curve. Useful for characterizing the classification accuracy of continuous measurements for predicting binary states

Usage

GeomRoc

geom_roc(
  mapping = NULL,
  data = NULL,
  stat = "roc",
  n.cuts = 10,
  arrow = NULL,
  lineend = "butt",
  linejoin = "round",
  linemitre = 1,
  linealpha = 1,
  pointalpha = 1,
  pointsize = 0.5,
  labels = TRUE,
  labelsize = 3.88,
  labelround = 1,
  na.rm = TRUE,
  cutoffs.at = NULL,
  cutoff.labels = NULL,
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  ...
)
GeomRoc

geom_roc(
  mapping = NULL,
  data = NULL,
  stat = "roc",
  n.cuts = 10,
  arrow = NULL,
  lineend = "butt",
  linejoin = "round",
  linemitre = 1,
  linealpha = 1,
  pointalpha = 1,
  pointsize = 0.5,
  labels = TRUE,
  labelsize = 3.88,
  labelround = 1,
  na.rm = TRUE,
  cutoffs.at = NULL,
  cutoff.labels = NULL,
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  ...
)

Arguments

`mapping`	Set of aesthetic mappings created by `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)`).
`stat`	Use to override the default connection between `geom_roc` and `stat_roc`.
`n.cuts`	Number of cutpoints to display along each curve
`arrow`	Arrow specification, as created by `arrow`
`lineend`	Line end style (round, butt, square)
`linejoin`	Line join style (round, mitre, bevel)
`linemitre`	Line mitre limit (number greater than 1)
`linealpha`	Alpha level for the lines, alpha.line is deprecated
`pointalpha`	Alpha level for the cutoff points, alpha.point is deprecated
`pointsize`	Size of cutoff points, size.point is deprecated
`labels`	Logical, display cutoff text labels
`labelsize`	Size of cutoff text labels
`labelround`	Integer, number of significant digits to round cutoff labels
`na.rm`	Remove missing values from curve
`cutoffs.at`	Vector of user supplied cutoffs to plot as points. If non-NULL, it will override the values of n.cuts and plot the observed cutoffs closest to the user-supplied ones.
`cutoff.labels`	vector of user-supplied labels for the cutoffs. Must be a character vector of the same length as cutoffs.at.
`position`	Position adjustment, either as a string naming the adjustment (e.g. `"jitter"` to use `position_jitter`), or the result of a call to a position adjustment function. Use the latter if you need to change the settings of the adjustment.
`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()`.
`...`	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.

Format

An object of class GeomRoc (inherits from Geom, ggproto, gg) of length 6.

Computed variables

false_positive_fraction: estimate of false positive fraction
true_positive_fraction: estimate of true positive fraction
cutoffs: values of m at which estimates are calculated

Aesthetics

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

x The FPF estimate. This is automatically mapped by stat_roc
y The TPF estimate. This is automatically mapped by stat_roc smallest level in sort order is assumed to be 0, with a warning
alpha
color
fill
linetype
size

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()

ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc()
ggplot(rocdata, aes(m = M, d = D)) + geom_roc() + facet_wrap(~ Z)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(n.cuts = 20)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(cutoffs.at = c(1.5, 1, .5, 0, -.5))
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(labels = FALSE)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(size = 1.25)

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()

ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc()
ggplot(rocdata, aes(m = M, d = D)) + geom_roc() + facet_wrap(~ Z)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(n.cuts = 20)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(cutoffs.at = c(1.5, 1, .5, 0, -.5))
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(labels = FALSE)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(size = 1.25)

Reads included JavaScript functions and returns them as a string for pasting into a webpage

Description

Reads included JavaScript functions and returns them as a string for pasting into a webpage

Usage

getD3()
getD3()

Plot an ROC curve

Description

Deprecated, use geom_roc instead

Usage

ggroc(
  rocdata,
  fpf_string = "FPF",
  tpf_string = "TPF",
  c_string = "c",
  ci = FALSE,
  label = NULL,
  label.adj.x = 0,
  label.adj.y = 0,
  label.angle = 45,
  plotmath = FALSE,
  xlabel = "False positive fraction",
  ylabel = "True positive fraction"
)
ggroc(
  rocdata,
  fpf_string = "FPF",
  tpf_string = "TPF",
  c_string = "c",
  ci = FALSE,
  label = NULL,
  label.adj.x = 0,
  label.adj.y = 0,
  label.angle = 45,
  plotmath = FALSE,
  xlabel = "False positive fraction",
  ylabel = "True positive fraction"
)

Arguments

`rocdata`	Data frame containing true and false positive fractions, and cutoff values
`fpf_string`	Column name identifying false positive fraction column
`tpf_string`	Column name identifying true positive fraction column
`c_string`	Column name identifying cutoff values
`ci`	Logical, not supported
`label`	Not supported
`label.adj.x`	Not supported
`label.adj.y`	Not supported
`label.angle`	Not supported
`plotmath`	Not supported
`xlabel`	Defaults to "False positive fraction"
`ylabel`	Defaults to "True positive fraction"

Value

A ggplot object

Transform biomarkers stored as wide to long

Description

Multiple biomarkers measured on the same subjects are often stored as multiple columns in a data frame. This is a convenience function that transforms the data into long format, suitable for use with ggplot and geom_roc

Usage

melt_roc(data, d, m, names = NULL)
melt_roc(data, d, m, names = NULL)

Arguments

`data`	Data frame containing disease status and biomarkers stored in columns
`d`	Column containing binary disease status. Can be a column name or index
`m`	Vector of column names or indices identifying biomarkers
`names`	Optional vector of names to assign to the biomarkers. If NULL, names will be taken from the column names

Value

A data frame in long format with three columns: D = binary disease status, M = biomarker value, and name = biomarker name

Examples

D.ex <- rbinom(50, 1, .5)
widedata <- data.frame(D = D.ex, M1 = rnorm(50, mean = D.ex, sd = 1), 
   M2 = rnorm(50, mean = D.ex, sd = .5))
longdata <- melt_roc(widedata, "D", c("M1", "M2"))
ggplot(longdata, aes(d = D, m = M, color = name)) + geom_roc()

D.ex <- rbinom(50, 1, .5)
widedata <- data.frame(D = D.ex, M1 = rnorm(50, mean = D.ex, sd = 1), 
   M2 = rnorm(50, mean = D.ex, sd = .5))
longdata <- melt_roc(widedata, "D", c("M1", "M2"))
ggplot(longdata, aes(d = D, m = M, color = name)) + geom_roc()

Plot multiple ROC curves

Description

Given a list of results computed by calculate_roc, plot the curve using ggplot with sensible defaults. Pass the resulting object and data to export_interactive_roc, plot_interactive_roc, or plot_journal_roc.

Usage

multi_ggroc(
  datalist,
  fpf_string = rep("FPF", length(datalist)),
  tpf_string = rep("TPF", length(datalist)),
  c_string = rep("c", length(datalist)),
  label = NULL,
  legend = TRUE,
  label.adj.x = rep(0, length(datalist)),
  label.adj.y = rep(0, length(datalist)),
  label.angle = rep(45, length(datalist)),
  plotmath = FALSE,
  xlabel = "False positive fraction",
  ylabel = "True positive fraction"
)
multi_ggroc(
  datalist,
  fpf_string = rep("FPF", length(datalist)),
  tpf_string = rep("TPF", length(datalist)),
  c_string = rep("c", length(datalist)),
  label = NULL,
  legend = TRUE,
  label.adj.x = rep(0, length(datalist)),
  label.adj.y = rep(0, length(datalist)),
  label.angle = rep(45, length(datalist)),
  plotmath = FALSE,
  xlabel = "False positive fraction",
  ylabel = "True positive fraction"
)

Arguments

`datalist`	List of data frames each containing true and false positive fractions and cutoffs
`fpf_string`	Column names identifying false positive fraction
`tpf_string`	Column names identifying true positive fraction
`c_string`	Column names identifying cutoff values
`label`	Not supported.
`legend`	If true, draws legend
`label.adj.x`	Not supported.
`label.adj.y`	Not supported.
`label.angle`	Not supported.
`plotmath`	Logical. Not supported.
`xlabel`	Defaults to "False positive fraction"
`ylabel`	Defaults to "True positive fraction"

Value

A ggplot object

Generate a standalone html document displaying an interactive ROC curve

Description

Generate a standalone html document displaying an interactive ROC curve

Usage

plot_interactive_roc(ggroc, file = NULL, ...)
plot_interactive_roc(ggroc, file = NULL, ...)

Arguments

`ggroc`	An object as returned by ggroc or multi_ggroc. It can be modified with annotations, themes, etc.
`file`	A path to save the result to. If NULL, will save to a temporary directory
`...`	arguments passed to export_interactive_roc

Value

NULL opens an interactive document in Rstudio or the default web browser

Plot an ROC curve for use in print

Description

Deprecated, use style_roc instead

Usage

plot_journal_roc(
  ggroc_p,
  font.size = 3,
  n.cuts = 20,
  ci.at = NULL,
  opacity = 0.3,
  lty = NULL,
  color = NULL,
  lwd = NULL,
  legend = FALSE
)
plot_journal_roc(
  ggroc_p,
  font.size = 3,
  n.cuts = 20,
  ci.at = NULL,
  opacity = 0.3,
  lty = NULL,
  color = NULL,
  lwd = NULL,
  legend = FALSE
)

Arguments

`ggroc_p`	An object as returned by ggroc or multi_ggroc. It can be modified with annotations, themes, etc.
`font.size`	Not supported
`n.cuts`	Not supported
`ci.at`	Not supported
`opacity`	Not supported
`lty`	Not supported
`color`	Not supported
`lwd`	Not supported
`legend`	Not supported

Value

A ggplot object

Tools for plotting ROC Curves

Description

Generate Useful ROC Curve Charts for Print and Interactive Use. This defines a set of stats and geoms for use with ggplot2. In addition, ggplot objects created with these geoms can be exported and turned into interactive plots for use on the web. The interactive features include hovering to display hidden labels, and clicking to reveal confidence regions.

Details

To get started, see geom_roc, geom_rocci, or the examples below. For transforming data, we also provide the convenience function melt_roc.

The vignette contains some examples, along with explanations of the results. To view, run vignette("examples", package = "plotROC")

Author(s)

Michael Sachs (@sachsmc)

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() + geom_rocci()

ggplot(rocdata, aes(m = M, d = D)) + geom_roc() + style_roc()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(sig.level = .01)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(n.cuts = 0) + 
geom_rocci(ci.at = quantile(rocdata$M, c(.1, .25, .5, .75, .9)))
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(linetype = 1)

rocplot <- ggplot(rocdata, aes(m = M, d = D)) + geom_roc()

plot_interactive_roc(rocplot)
plot_interactive_roc(rocplot + aes(color = Z))
plot_interactive_roc(rocplot + facet_wrap( ~ Z))


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() + geom_rocci()

ggplot(rocdata, aes(m = M, d = D)) + geom_roc() + style_roc()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci()
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(sig.level = .01)
ggplot(rocdata, aes(m = M, d = D)) + geom_roc(n.cuts = 0) + 
geom_rocci(ci.at = quantile(rocdata$M, c(.1, .25, .5, .75, .9)))
ggplot(rocdata, aes(m = M, d = D, color = Z)) + geom_roc() + geom_rocci(linetype = 1)

rocplot <- ggplot(rocdata, aes(m = M, d = D)) + geom_roc()

plot_interactive_roc(rocplot)
plot_interactive_roc(rocplot + aes(color = Z))
plot_interactive_roc(rocplot + facet_wrap( ~ Z))

Key for ROC geom

Description

Key for ROC geom

Usage

roc_key(data, params, size)
roc_key(data, params, size)

Arguments

`data`	Data created by stat
`params`	parameters
`size`	Size

Start the plotROC Shiny app

Description

A convenience function to easily start the shiny application. It will open in Rstudio, or in the default web browser.

Usage

shiny_plotROC()
shiny_plotROC()

Calculate the empirical Receiver Operating Characteristic curve

Description

Given a binary outcome d and continuous measurement m, computes the empirical ROC curve for assessing the classification accuracy of m

Usage

StatRoc

stat_roc(
  mapping = NULL,
  data = NULL,
  geom = "roc",
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  na.rm = TRUE,
  max.num.points = 1000,
  increasing = TRUE,
  ...
)
StatRoc

stat_roc(
  mapping = NULL,
  data = NULL,
  geom = "roc",
  position = "identity",
  show.legend = NA,
  inherit.aes = TRUE,
  na.rm = TRUE,
  max.num.points = 1000,
  increasing = TRUE,
  ...
)

Arguments

`mapping`	Set of aesthetic mappings created by `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 to display the data, either as a `ggproto` `Geom` subclass or as a string naming the geom stripped of the `geom_` prefix (e.g. `"point"` rather than `"geom_point"`)
`position`	Position adjustment, either as a string naming the adjustment (e.g. `"jitter"` to use `position_jitter`), or the result of a call to a position adjustment function. Use the latter if you need to change the settings of the adjustment.
`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()`.
`na.rm`	Remove missing observations
`max.num.points`	maximum number of points to plot
`increasing`	TRUE (default) if M is positively associated with Pr(D = 1), if FALSE, assumes M is negatively associated with Pr(D = 1)
`...`	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.

Format

An object of class StatRoc (inherits from Stat, ggproto, gg) of length 6.

Aesthetics

stat_roc 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 Controls the label alpha, see also linealpha and pointalpha
color
linetype
size Controls the line weight, see also pointsize and labelsize

Computed variables

false_positive_fraction: estimate of false positive fraction
true_positive_fraction: estimate of true positive fraction
cutoffs: values of m at which estimates are calculated

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)) + stat_roc()
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)) + stat_roc()

Calculate confidence regions for the empirical ROC curve

Description

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,
  ...
)
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 `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 to display the data, either as a `ggproto` `Geom` subclass or as a string naming the geom stripped of the `geom_` prefix (e.g. `"point"` rather than `"geom_point"`)
`position`	Position adjustment, either as a string naming the adjustment (e.g. `"jitter"` to use `position_jitter`), or the result of a call to a position adjustment function. Use the latter if you need to change the settings of the adjustment.
`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.

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)))

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)))

Add guides and annotations to a ROC plot

Description

Adds a diagonal guideline, minor grid lines, and optionally direct labels to ggplot objects containing a geom_roc layer.

Usage

style_roc(
  major.breaks = c(0, 0.1, 0.25, 0.5, 0.75, 0.9, 1),
  minor.breaks = c(seq(0, 0.1, by = 0.01), seq(0.9, 1, by = 0.01)),
  guide = TRUE,
  xlab = "False positive fraction",
  ylab = "True positive fraction",
  theme = theme_bw
)
style_roc(
  major.breaks = c(0, 0.1, 0.25, 0.5, 0.75, 0.9, 1),
  minor.breaks = c(seq(0, 0.1, by = 0.01), seq(0.9, 1, by = 0.01)),
  guide = TRUE,
  xlab = "False positive fraction",
  ylab = "True positive fraction",
  theme = theme_bw
)

Arguments

`major.breaks`	vector of breakpoints for the major gridlines and axes
`minor.breaks`	vector of breakpoints for the minor gridlines and axes
`guide`	logical, if TRUE draws diagonal guideline
`xlab`	X-axis label
`ylab`	Y-axis label
`theme`	Theme function compatible with ggplot2

Examples

D.ex <- rbinom(50, 1, .5)
fakedata <- data.frame(M1 = rnorm(50, mean = D.ex), 
   D = D.ex)
ggplot(fakedata, aes(m = M1, d = D)) + geom_roc() + style_roc()
ggplot(fakedata, aes(m = M1, d = D)) + geom_roc() + style_roc(xlab = "1 - Specificity")
ggplot(fakedata, aes(m = M1, d = D)) + geom_roc() + style_roc(theme = theme_grey)
D.ex <- rbinom(50, 1, .5)
fakedata <- data.frame(M1 = rnorm(50, mean = D.ex), 
   D = D.ex)
ggplot(fakedata, aes(m = M1, d = D)) + geom_roc() + style_roc()
ggplot(fakedata, aes(m = M1, d = D)) + geom_roc() + style_roc(xlab = "1 - Specificity")
ggplot(fakedata, aes(m = M1, d = D)) + geom_roc() + style_roc(theme = theme_grey)

Check that D is suitable for using as binary disease status

Description

Checks for two classes and gives a warning message indicating which level is assumed to be 0/1. Throws an error if more than two levels appear in D.

Usage

verify_d(D)
verify_d(D)

Arguments

`D`	Vector that will be checked for 2-class labels

Value

A vector the same length as D that takes values 0, indicating no disease or 1 indicating disease.

Examples


verify_d(c(1, 0, 1))
## Not run: 
verify_d(c(TRUE, FALSE, TRUE)) #warning
verify_d(c("Dead", "Alive", "Dead")) #warning
verify_d(c("Disease", "Healthy", "Missing")) #error

## End(Not run)

verify_d(c(1, 0, 1))
## Not run: 
verify_d(c(TRUE, FALSE, TRUE)) #warning
verify_d(c("Dead", "Alive", "Dead")) #warning
verify_d(c("Disease", "Healthy", "Missing")) #error

## End(Not run)

Package 'plotROC'

Help Index

Calculate the Area under the ROC curve

Description

Usage

Arguments

Value

Examples

Calculate the Empirical ROC curves for multiple biomarkers

Description

Usage

Arguments

Value

Calculate the Empirical ROC curve

Description

Usage

Arguments

Details

Value

Add direct labels to a ROC plot

Description

Usage

Arguments

Generate svg code for an ROC curve object

Description

Usage

Arguments

Details

Value

Confidence regions for the ROC curve

Description

Usage

Arguments

Format

Aesthetics

See Also

Examples

Empirical Receiver Operating Characteristic Curve

Description

Usage

Arguments

Format

Computed variables

Aesthetics

See Also

Examples

Reads included JavaScript functions and returns them as a string for pasting into a webpage

Description

Usage

Plot an ROC curve

Description

Usage

Arguments

Value

Transform biomarkers stored as wide to long

Description

Usage

Arguments

Value

Examples

Plot multiple ROC curves

Description

Usage

Arguments

Value

Generate a standalone html document displaying an interactive ROC curve

Description

Usage

Arguments

Value

Plot an ROC curve for use in print

Description

Usage

Arguments

Value

Tools for plotting ROC Curves

Description

Details

Author(s)

See Also