ROC.default {sensR} | R Documentation |
The function computes and plots the empirical ROC (receiver operating characteristic) curve.
ROC(object, ...) ## Default S3 method: ROC(object, se.d, scale = 1, length = 1000, fig = TRUE, se.type = c("CI", "SE"), CI.alpha = 0.05, ...) ## S3 method for class 'discrim': ROC(object, length = 1000, fig = TRUE, se.type = c("CI", "SE"), CI.alpha = 0.05, ...)
object |
the class of the object defines, which of the methods is
invoked. If obejct is a single element numeric vector it is taken as
a d-prime value and the default method is invoked. If the object is
of class discrim (works for AnotA objects), the method for discrim
objects is invoked. |
se.d |
a unit length vector with the standard error of d-prime. If supplied confidence intervals or standard errors are plotted |
scale |
a unit length vector giving the ratio of scale (ie. standard deviation) of the latent distribution for the no-class items relative to that of the yes-class items |
length |
the length of the vectors to be plotted. Longer vectors gives more smooth curves. |
fig |
Should a plot be produced? |
se.type |
The type of band for the ROC curve, "CI" for
confidence interval and "SE" for standard error. |
CI.alpha |
the type I level of the confidence interval of AUC |
... |
additional arguments to plot and lines |
The function currently ignores the variance of the scale in the computation of the uncertainty of the ROC curve.
The function makes a plot of the ROC curve, and if se.d
is
supplied, standard errors or confidence intervals for the curve are
added to the plot.
The function also (invisibly) returns a list with the following
components
ROCx |
x-coordinates to the ROC curve |
ROCy |
y-coordinates to the ROC curve |
lower |
y-coordinates to the lower limit |
upper |
y-coordinates to the upper limit |
Rune Haubo B Christensen
## ROC.default: (mat <- matrix(c(8, 17, 1, 24), 2, byrow = TRUE)) (d.prime <- SDT(mat, "probit")[3]) ROC(d.prime) ## ROC.discrim: fm1 <- AnotA(8, 25, 1, 25) ROC(fm1)