Utility-Based Accuracy Measures to Empirically Evaluate Credal Classifiers

Predictions made by imprecise-probability models are often indeterminate (that is, set-valued). Measuring the quality of an indeterminate prediction by a single number is important to fairly compare different models, but a principled approach to this problem is currently missing. In this paper we derive a measure to evaluate the predictions of credal classifiers from a set of assumptions. The measure turns out to be made of an objective component, and another that is related to the decision-maker’s degree of risk-aversion. We discuss when the measure can be rendered independent of such a degree, and provide insights as to how the comparison of classifiers based on the new measure change with the number of predictions to be made. Finally, we empirically study the behavior of the proposed measure.

1