Carding machine learning basis - (accuracy, precision, recall Discussion)

A, TP TN FP FN

TP: label positive example, the prediction is positive cases (P), i.e., correctly predicted (T)

TN: negative label embodiment, the prediction is negative cases (N), i.e., correctly predicted (T)

FP: negative label embodiment, the prediction is positive cases (P), i.e., the prediction error (F)

FN: Label a positive example, the prediction is negative example (N), i.e., the prediction error (F)

Wherein T: True F: False P: Positive N: Negative

As the acronym is more difficult to remember, I will be respectively referred to as: true positive samples (TP), true negative samples (TN), false positive samples (FP), false negative samples (FN)

 

二、accuracy precision recall

Accuracy: accuracy = (TP + TN) / (TP + TN + FP + FN), that is to predict the correct sample proportion accounted for all samples

Accuracy: precision = TP / (TP + FP), i.e. the proportion of total true positive samples predicted positive samples

Recall: recall = TP / (TP + FN), that is, all positive samples correctly predicted how many

Guess you like

Origin www.cnblogs.com/cookbook/p/12310262.html