ABSTRACT
Basic ROC curves summarize the score distributions of objects from two classes. Situations requiring such summaries are very commonand hence the widespread use of ROC curves and the statistics derived from them. However, in many situations things are more complicated than this basic notion. Clearly such further complications can arise in an unlimited number of ways, so we cannot examine all of them. Instead, in this chapter, we describe some of the work on just a few of the most important such generalizations.
