分类指标

首先写出confusion matrix。

常见指标

accuracy = (TP + TN) / (TP + TN + FN + FP)

precision = TP / (TP + FP)

recall = TP / (TP + FN)

误报率 = false alarm rate = false positive rate = FP / (FP + TN)

漏报率 = false negative rate = FN / (TP + FN)

灵敏度 = sensitivity = true positive rate = recall = TP / (TP + FN)

特异度 = specificity = false positive rate = 误报率 = FP / (FP + TN)

ROC曲线

ROC曲线的纵轴是true positive rate，纵轴是false positive rate。曲线集中在左上角比较好一点。

PR曲线

PR曲线的纵轴是precision，横轴是recall。曲线集中在右上角比较好一点。