Auc(曲线下面积)作为分类器性能的评价指标,得到了广泛的应用.但其存在一定的缺陷.讨论AUC和总误差之间的关系,从而证明了AUC的不一致性.
The area under the receiver operating characteristic curve (AUC) is a widely used classification perform- ance measure in machine learning. However, there are some disadvantages. In this paper, the relationship between AUC and the total errors is discussed, and the incoherence of AUC is showed.