Nice Negra Y Educada Halloween Shirt
When working with imbalanced data, you would normally use true and false-positive rates. One measures the odds someone has COVID given that they tested positive and the other is the odds that they don’t give they tested positive. What you then do is adjust the sensitivity of the test (just about every type of test or statistical model has a way to adjust sensitivity) to increase and decrease the TPR and FPR. There is a trade-off between the two and you can plot this relationship on a graph to get a curve called the receiver operating characteristic. The metric to actually pay attention to is the area under that curve (AUC or AUROC).
Nhận xét
Đăng nhận xét