eer: Compute the equal error rate given false positive and false...

In davidavdav/ROC: Compute structures to compute ROC and DET plots and metrics for 2-class classifiers.

Compute the equal error rate given false positive and false negative rates, using the convex hull method.

Description

This routine interpolates the false positive and false negative rates on the convex hull of the ROC curve, to find the point where they are equal, and returns this value. This is a convenience function for people who are only interesed in a (consistent) EER value without going through the hassle of obtaining a roc object.

Usage

  eer(pfa, pmissm index=NULL)

Arguments

pfa	probability of false alarm, or false positive rate
pmiss	probability of miss, or false negarive rate. Also known as the complement of the hit rate, or one minus the true positive rate.
index	a vector of indices indicating which points ly on the convex hull, or NULL to indicate the convex hull is to be computed by eer().

Value

This value returns the interpolated value where the pfa==pmiss, for points lying on the convex hull of the ROC.

See Also

Cllr.tnt, pAUC

Examples

data(ru.2009)
x <- subset(ru.2009, mcond=="TC6" & tcond=="TS2")
r <- roc(x)
summary(r)
eer(r$pfa, r$pmiss, which(r$chull))
## re-compute the EER without the convex-hull intermpretation
eer(r$pfa, r$pmiss, 1:nrow(r))

davidavdav/ROC documentation built on Sept. 8, 2023, 2:39 p.m.