During the evaluation of the Target Detection results, the terms false positve, precision rate, and recall rate are always required. I always know the meaning of these words when I use them, but I forgot it for a long time. I discussed this issue with my senior friend a few days ago. I decided to simply summarize and write it down for future use.
The chinese meanings of these three words are: false detection rate (also known as false alarm rate), precision rate and total query rate. You can see how they are calculated:
To understand the meaning of TP, FP, FN, and TN, it is basically clear. False positve is widely used in medical testing. We will use a tumor as an example:
A patient with a tumor goes to the hospital for examination and has the following situations:
1. Actual Situation: malignant. Check Result: malignant. This situation is called True positve. (Unlucky enough)
2. Actual Situation: benign. Check Result: malignant. This situation is called false positve and false positive. (No disease, no reason)
3. Actual Situation: malignant. Check Result: benign. This situation is called false negative, false negative. (Yongzhi is delaying his illness)
4. Actual Situation: benign. Check Result: benign. This situation is called True postive, which is actually negative. (This guy has a good character)
After understanding these terms, we can know how the three "rates" are calculated:
False check rate: FP rate = sum (FP)/(sum (FP) + sum (TN ))
Precision: precision rate = sum (TP)/(sum (TP) + sum (FP ))
Query rate: recall rate = sum (TP)/(sum (TP) + sum (FN ))
Another common missed detection rate is: Miss rate = sum (FN)/(sum (TP) + sum (FN ))
Recall rate + miss rate = 1
The false check rate is the number of false targets identified as the actual target. The precision rate is the proportion of the detected targets; the check-out rate is the ratio of the detected targets to all targets.
Well, I don't know if you have understood it. I am dizzy every time. Good luck.