groc: ROC curve

In R4VN: Health Data Analysis and Publication-Ready Reporting

ROC curve

Description

Calculates and draws a receiver operating characteristic curve from a binary outcome and numeric predicted probabilities or scores. No external package is required.

Usage

groc(
  data = NULL, outcome, pred = NULL, vars = NULL, by = NULL, event = NULL,
  diagonal = TRUE, auc = TRUE, digits = 3, curve_lty = 1, curve_lwd = 2.5,
  diagonal_color = "gray60", diagonal_lty = 2, diagonal_lwd = 1, xlab = NULL,
  ylab = NULL, xtitle = "1 - Specificity", ytitle = "Sensitivity",
  title = NULL, subtitle = NULL, note = NULL, color = NULL,
  palette = "journal", xline = NULL, yline = NULL, ref_color = "gray40",
  ref_lty = 2, ref_lwd = 1, theme = "journal", size = 11, combine = FALSE,
  ncol = NULL, file = NULL, width = 6, height = 6, dpi = 300, show = TRUE,
  bg = "white", vline = NULL, hline = NULL
)

Arguments

data	A data frame.
outcome	Binary outcome variable.
pred	Numeric predicted probability or score; larger values must indicate a greater probability of the event.
vars	Optional vars(...) selector for several numeric predictors or scores.
by	Optional grouping variable or hierarchical vars(...) selector.
event	Event value. By default, the second factor level or the larger numeric value is used.
diagonal	Show the no-discrimination diagonal.
auc	Show the area under the curve.
digits	Number of AUC digits.
curve_lty, curve_lwd	ROC-curve line type and width.
diagonal_color, diagonal_lty, diagonal_lwd	Colour, line type, and width of the no-discrimination diagonal.
xlab, ylab	Optional tick labels.
xtitle, ytitle	Axis titles.
title, subtitle, note, color, palette, theme, size, combine, ncol, file, width, height, dpi, show, bg	See gbar().
xline, yline	Optional numeric reference lines on the x and y axes.
ref_color, ref_lty, ref_lwd	Color, line type, and width for reference lines.
vline, hline	Deprecated aliases for xline and yline.

Details

Draws ROC curves for one or several predictor/score variables. by can create ROC analyses within hierarchical strata.

Value

An object of class r4vn_graph; its data component contains thresholds, sensitivity, and specificity, and its auc attribute contains the AUC.

Examples

d <- data.frame(y = c(0, 0, 1, 1, 1), p = c(.10, .35, .40, .75, .90))
groc(d, outcome = y, pred = p, event = 1)

# Extended usage examples

d <- data.frame(
  outcome = factor(c("No", "No", "No", "Yes", "Yes", "Yes", "Yes")),
  probability = c(.05, .20, .35, .40, .65, .80, .95)
)

# ROC curve with explicit event
roc1 <- groc(d, outcome, probability, event = "Yes", show = FALSE)
attr(roc1$data, "auc")

# Hide the diagonal or AUC annotation and customize labels
groc(d, outcome, probability, event = "Yes",
     diagonal = FALSE, auc = FALSE,
     xtitle = "False-positive rate", ytitle = "True-positive rate",
     show = FALSE)

# Numeric binary outcome; the larger value is the default event
d2 <- data.frame(y = c(0, 0, 1, 1, 1), score = c(.10, .35, .40, .75, .90))
groc(d2, y, score, show = FALSE)

# Export the ROC curve
groc(d, outcome, probability, event = "Yes",
     file = tempfile(fileext = ".pdf"), show = FALSE)

R4VN documentation built on Sept. 30, 2026, 5:13 p.m.