Calculates the cross-validated AUC (concordance) from a cv.ncvsurv
object.
# S3 method for cv.ncvsurv
AUC(obj, ...)
A cv.ncvsurv
object. You must run cv.ncvsurv()
with the
option returnY=TRUE
in order for AUC()
to work
For S3 method compatibility; not used
The area under the curve (AUC), or equivalently, the concordance statistic
(C), is calculated according to the procedure described in van Houwelingen
and Putter (2011). The function calls survival::concordancefit()
, except
except cross-validated linear predictors are used to guard against
overfitting. Thus, the values returned by AUC.cv.ncvsurv()
will be lower
than those you would obtain with concordancefit()
if you fit the full
(unpenalized) model.
van Houwelingen H, Putter H (2011). Dynamic Prediction in Clinical Survival Analysis. CRC Press.
data(Lung)
X <- Lung$X
y <- Lung$y
cvfit <- cv.ncvsurv(X, y, returnY=TRUE)
head(AUC(cvfit))
#> [1] 0.5409473 0.6172194 0.6607224 0.6759428 0.6842344 0.6838937
lam <- cvfit$lambda
plot(lam, AUC(cvfit), xlim=rev(range(lam)), lwd=3, type='l',
las=1, xlab=expression(lambda), ylab='AUC')