当前位置：网站首页>[indicators] precision, recall

[indicators] precision, recall

2022-04-23 13:18:00 【^_^ Min Fei】

List of articles

1. Binary classification index
2. Reference resources

1. Binary classification index

Insert picture description here
TP： Forecast as Real sample in Indeed Really quantity
FP： Forecast as Real sample in Indeed Number of fake
FN： Forecast as Fake samples in Indeed Really quantity
TN： Forecast as Fake samples in Indeed Number of fake

Accuracy rate Accuracy：

$\frac{ Correctly predicted samples = The correct prediction is 1+ The correct prediction is 0 Of }{ The total number of samples }$ = $\frac{TP + TN}{ALL}$

Precision rate Precision：

$\frac{ The number of true samples in which all predictions are true }{ The number of samples predicted to be true }$ = $\frac{TP}{TP + FP}$

Recall rate Recall：

$\frac{ The number of true samples in which all predictions are true }{ Number of all true samples }$ = $\frac{TP}{TP + FN}$

give an example ： Predict the number of diseases .
Suppose there are 100 Samples , among 20 A sick man . Through prediction, it is found that 24 A sick man , Among them is 18 One is really sick .
In this case ：
TP = 18
FP = 24-18=6
FN= 20-18 =2
TN= （100-24）-2 = 74

Accuracy rate = $\frac{18+74}{100}=0.82$
Precision rate = $\frac{18}{18+6}$
Recall rate = $\frac{18}{18+2}$

2. Reference resources

https://www.jianshu.com/p/1ca37d729d0c

版权声明
本文为[^_^ Min Fei]所创，转载请带上原文链接，感谢
https://yzsam.com/2022/04/202204230611136755.html