Analyzing Receiver Operating Characteristic Curves with SAS

Single Continuous Predictor

3.1 Dichotomizing a Continuous Predictor

3.2 The ROC Curve

3.3 Empirical ROC Curve and the Conditional Distributions of the Marker

3.4 Area under the ROC Curve

3.5 Selecting an Optimal Threshold

3.6 The Binormal ROC Curve

3.7 Transformations to Binormality

3.8 Direct Estimation of the Binormal ROC Curve

3.9 Bootstrap Confidence Intervals for the Area under the Curve

3.1 Dichotomizing a Continuous Predictor

Now let's consider the problem that naturally leads to the use of the ROC curve. Suppose you are trying to predict a binary outcome, like the weather forecast, but instead of a binary predictor you have a continuous ...