… , an average of aggregate measures, which is different from … [11] DJ Hand, “Measuring
classifier performance: A coherent … index for evaluating the classification performance at a high …

… the context or that interpreting the AUC of intersecting ROC … these aggregated results for
all of the performance metrics. … is that the ROC AUC and H-Measure metrics, while not being …

… In this work, we explore AUC as a performance measure in … Even though the ROC graph has
visual appeal, an aggregated … Although classification scores may not be interpreted as strict …

… of aggregation and carefully delineates the interpretations of each in … AUCs, as a higher
AUC does not imply an everywhere dominating ROC curve. As a scalar measure of performance…

… For example, classification performance can be measured by … (AUC) is itself meaningful as
an aggregated performance … without checking that it is indeed coherent to do so. These issues …

… role, being a popular performance measure when a singular … by checking both the PR AUC
and the ROC AUC, focusing on … the only aggregate rate highlighting the weak performance of …

… if only 3 classifier results are aggregated. Moreover, our results reveal … The H -measure has
been developed as a coherent … are very similar to those generated with the AUC measure. …

… computed the area under curve (AUC) values. Additionally, we … We also compared performance
of the deep learning models … We then aggregated disease probability maps across the …

… Multi-level classification aggregates the … of AUC and H measure as well. Similar to what
was observed for the Australian dataset, the classifier showing relatively superior performance …

… images of retinal optical coherence tomography (OCT) with 4 … We also observe that the
ensemble improves the AUC only … the ensemble’s performance by aggregating the models’ …