There are a number of words used very frequently to describe properties of a confusion matrix (or contingency table).

Among them there is the word 'recall', this is to describe:


which means, the rate of your predicted positive instance out of the real true instance number (the true positives plus false negatives).

But this was given the name recall, which has confused me a lot.

After some digging within Cambridge Dictionary, I found that 'recall' sounds to me like you have your memory back and then you think of something you might have forgotten. Another is to describe how you want something you already sent out back.

So why can recall the word be used to describe the confusion matrix rate? When (and by who) was the first use of 'recall' in this way?

  • 1
    Hello, cinqS. Please include the research you’ve done, such as looking up 'recall' in the dictionary to see if there are alternative meanings. Questions that can be answered using commonly-available references are off-topic. Until reasonable research is submitted, this seems a reasonable assumption for users to make. – Edwin Ashworth Jan 29 '18 at 12:06
  • @EdwinAshworth, Hi, yes, I did search some pages, and already updated the question I made. thanks for your notice. – cinqS Jan 29 '18 at 12:21
  • 1
    This site has not implemented MathJax. Please do your best with ordinary Markdown. However, I suspect that this might be a better fit on Statistics (Cross Validated) or maybe even History of Science and Mathematics. – Andrew Leach Jan 29 '18 at 13:19
  • It's idiosyncratic to the context of classifiers, referring to the completeness of results given. Think of it as something like sending out the test cases and retrieving or recalling a certain percentage. You'd be better off putting this in a stats board though, because it's more about a context sensitive analogy, and somewhere that can do a much better job. Also see en.wikipedia.org/wiki/Precision_and_recall. – jimm101 Feb 1 '18 at 12:16
  • This would be a good general etymology question, who first started using 'recall' in this way (likewise for all the other terms, sensitivity, specificity, precision, etc etc, because none of their everyday meanings are obviously connected with their mathematical meanings. But yeah probably more likely to get info from someone on Cross Validated or HSM as @AndrewLeach mentions, people who are knowledgable about the history of those areas. – Mitch Aug 1 at 15:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.