biasproproc: Bias in Using AUC of a Proper ROC Curve for Diagnostic...

In parodistefano/properROC: Display and Analyze Proper ROC Curves

Description

Estimate the difference between AUC and the global diagnostic accuracy, i.e., the proportion of corrected classifications expected at the best cut-off of a proper ROC curve in a Proper ROC Model

Usage

biasproproc(dor)

Arguments

dor	Diagnostic Odds Ratio (the parameter of the expected proper curve)

Value

the bias, the difference between AUC and the "best" expected accuracy

Examples

ROCCa125 <- roctable(PancreaticData$Ca125, PancreaticData$Status)
BiasCa125 <- biasproproc(rocdor(rocauc(ROCCa125)))
print(paste("Bias for the proper ROC curve for Ca125 tumor marker: ",
      round(BiasCa125,4)))

parodistefano/properROC documentation built on May 24, 2019, 6:16 p.m.