s.roc: Get ROC Curve Data for Binary Classification

In ldt: Automated Uncertainty Analysis

Get ROC Curve Data for Binary Classification

Description

This function calculates the required points for plotting the ROC curve and the AUC.

Usage

s.roc(y, scores, weights = NULL, options = get.options.roc())

Arguments

y	A numeric vector (Nx1) representing the actual values.
scores	A numeric vector (Nx1) representing the calculated probabilities for the negative observations.
weights	A numeric vector (Nx1) representing the weights of the observations. Use NULL for equal weights.
options	A list from get.options.roc function for more options.

Details

This is generally a statistics helper method in this package and it shows the inner calculations. See AUC section in get.search.metrics for a discussion.

Value

A list with the following items:

n	Number of observations.
auc	Value of AUC.
points	Points for plotting ROC.

Examples

y <- c(1, 0, 1, 0, 1, 1, 0, 0, 1, 0)
scores <- c(0.1, 0.2, 0.3, 0.5, 0.5, 0.5, 0.7, 0.8, 0.9, 1)
res1 <- s.roc(y,scores)
costs <- c(1,2,1,4,1,5,1,1,0.5,1)
costMatrix <- matrix(c(0.02,-1,-3,3),2,2)
opt <- get.options.roc(costs = costs, costMatrix = costMatrix)
res2 <- s.roc(y,scores,NULL,options = opt)

ldt documentation built on Sept. 11, 2024, 5:25 p.m.