[R 통계분석] 오차 행렬 (Confusion Matrix)

출처 : https://blog.naver.com/sjy5448/222458248562

 

Sensitivity 민감도 = TP/(TP+FN)

Specificity 특이성 = TN/(TN+FP)

Accuracy 정확도 = (TPs + TNs)/(Ts+Ns)

 

예제 MTE: Junk Food Tester

You are working for a school as a data scientist and your job is to make an AI model for lunch menus called Junk Food Tester (JFT), which will generate an alarm for Junk Food. JFT will scan each image of food and will generate either + or - result.

 

Here is the result after scanning 500 food images

- There are 100 Junk Food and 400 non-Junk Food images.

- From 100 Junk Food images, 90 were classified as Junk Food.

- From 400 non-Junk Food images, 100 were classified as Junk Food.

 

TP: 90FP: 10TN: 300

FN: 100

 

Sensitivity = 90/(90+100) = 90/190 = 0.47

Specificity = 300/(300+10) = 300/310 = 0.97

Accuracy = (Sensitivity+Specificity)/2 = 0.72 혹은 (TPs+TNs)/(Ts+Ns) = (90+300)/500 = 0.78

 

 

 

 

 

+ 추가내용

Predictive Value

Positive Predictive Value (PPV, PV+): the likelihood that the prediction is correct if the test result is positive.

Negative Predictive Value (NPV, PV-): the likelihood that the prediction is correct if the test result is negative.