Research overview
Solving the inverse ROC problem from finite samples.
The thesis examines how to estimate the decision threshold associated with a target operating point on a receiver operating characteristic curve. The method is nonparametric: it does not require a predefined distribution for either class.
It develops the estimator's asymptotic distribution and uses simulation to study its behavior with finite samples, providing a foundation for threshold selection in diagnostic testing and binary classification.
ROC curveThreshold estimationNonparametric statisticsBinary classificationDiagnostic testingStatistics
Suggested citation: Yousef, Waleed Ahmed. “Nonparametric Estimation of the Threshold at an Operating Point on the ROC Curve.” M.Sc. thesis, The George Washington University, 2007.