精确率(precision)或阳性预测值(Positive Predictive Value, PPV):判断正确的结果占预测为positive的比例, P r e c i s i o n = T P T P + F P Precision= \frac{TP}{TP+FP} Precision=TP+FPTP
灵敏度(Sensitivity)或称召回率(Recall)或真阳率(True Positive Rate, TPR):模型预测正确的个数占真实值为positive的比例, S e n s i t i v i t y = T P T P + F N Sensitivity= \frac{TP}{TP+FN} Sensitivity=TP+FNTP
特异度(Specificity)或选择率(Selectivity)或真阴率(True Negative Rate, TNR):判断正确的个数占真实值为Negative的比例, S p e c i f i c i t y = T N T N + F P Specificity= \frac{TN}{TN+FP} Specificity=TN+FPTN
阴性预测值(Negative Predictive Value, NPV):判断正确的结果占预测为negative的比例, N P V = T N T N + F N NPV= \frac{TN}{TN+FN} NPV=TN+FNTN
假阴率(False Negative Rate, FNR):判断错误的个数占真实值为positive的比例, F N R = F N F N + T P FNR= \frac{FN}{FN+TP} FNR=FN+TPFN
假阳率(False Positive Rate, F