ROC Flexibility Data 

The ROC Flexibility data is an artificially derived dataset with 5 features. It has some properties that make it particularly useful for testing supervised ranking algorithms. Namely, linear combinations of the 5 features can yield a wide variety of ranked lists, including some that have larger area-under-the-curve (AUC) values, and some lists that have larger values of other rank statistics that concentrate at the top of the list. However, there is no linear combination that can give an almost perfect list, which means that optimizing one rank statistic does not imply a good value on another one.

ROC Flexibility Features

ROC Flexibility Labels