roc_cut: Optimal cut for ROC curve
In jedwards24/jemodel: Personal package to help with modelling

roc_cut

R Documentation

Optimal cut for ROC curve

Description

Calculates the "optimal" cutoff for a classifier from a ROC curve. This is the cutoff that minimises the distance from the point (FPR = 0, TPR = 1). Also returns sensitivity and specitivity for this cutoff, the AUC of the ROC curve, and optionally plots the curve together with the distance measure.

Usage

roc_cut(pred, actual, plot = T)

Arguments

`pred`	A vector of prediction probabilities.
`actual`	A vector of outcome classes corresponding to `pred`.
`plot`	Logical indicating whether to plot a ROC curve together with distance from the optimal corner.