Calculates the area under the curve given a function or a fitted model.
1 2 3 4 5 6 |
x |
a function with range and domain in ROC space (default method) or an object of class |
fpr |
numeric vector, points on which the (S)ROC curve is evaluated |
level |
numeric, confidence level for the calculations of confidence intervals. |
sroc.type |
character, which SROC curve should be used to calculate the AUC? Besides the default |
... |
further arguments, currently not used. |
The area under the curve is calculated using the trapezoidal rule. The argument fpr
is the grid on which the (S)ROC curve is evaluated. In many cases the default grid will contain points on which the SROC curve can only be calculated by extrapolation; however if only a subinterval is specified a partial AUC is calculated and the AUC value might differ substantially.
For phm
objects the AUC and its confidence interval is calculated analytically, for reitsma
objects a call to the default method is invoked.
An object of the class AUC
which is really a list with component AUC
and an optional component ci
, which is currently only available from the AUC
method for phm
ojects.
Philipp Doebler <philipp.doebler@googlemail.com>
1 2 |
Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.
All documentation is copyright its authors; we didn't write any of that.