10
$\begingroup$

precision is the fraction of retrieved instances that are relevant, while recall (also known as sensitivity) is the fraction of relevant instances that are retrieved.

I know their meaning but I don't know why it is called recall? I am not a native-speaker of English. I know recall means remember, then I don't know the relevance of this meaning to this concept! maybe coverage was better because it shows how many instances were covered...or any other term.

Moreover sensitivity is also insensible to me!

Could you please help me to associate these words to the concept and have a sense of them?

$\endgroup$
  • $\begingroup$ "precision is the fraction of retrieved instances that are relevant, while recall (also known as sensitivity) is the fraction of relevant instances that are retrieved." this definition sounds quite concise. Where did you get it from? $\endgroup$ – Sanghyun Lee Dec 12 '19 at 21:19
4
$\begingroup$

I think the term "sensitivity" comes from the world of medical tests. A very sensitive test will test positive for most or all people who take the test and really have a disease, as well as for many people who don't. This corresponds to high recall, which means the query retrieves most or all of the relevant documents, as well as many that may not be relevant.

$\endgroup$
3
$\begingroup$

Recall means to bring back or remember. The terminology comes from information retrieval where it's usually being applied to a result set from a query. I suppose the sense of it is, how much of the set of right answers was retrieved by the query? how much of it was recalled?

I don't know if "coverage" is better or not. The word "sensitivity" is also used to mean the same thing as "recall". In any event, these are just the standard words for these ideas.

$\endgroup$
  • 1
    $\begingroup$ And another one - true positive rate. $\endgroup$ – stmax Jun 17 '16 at 20:13
  • 1
    $\begingroup$ Thank you, then its sense is close to "retrieve", "retrieve" itself was better! $\endgroup$ – Ahmad Jun 18 '16 at 4:55
  • $\begingroup$ I modified my question a bit. $\endgroup$ – Ahmad Jun 18 '16 at 6:08
2
$\begingroup$

recall is called 'recall' because it's the fraction of relevant (training-set) instances which were 'recalled' (yes or 'retrieved' as you suggest. 'coverage' would be more ambiguous, it could be misinterpreted as other things, e.g. the % of the training-set that you trained on (e.g. a partition. Your suggested 'coverage' would make sense wrt training-set but not the test-set, hence it's way too ambiguous).

I always assumed the reason 'recall' was also known as 'sensitivity' was due to signal-processing or medical influences: think a mine-detector, or a radar, or a test for a disease: 'sensitive' would mean it picks up most/all of the known relevant exemplars (from training-set).

Yes, most of these terms' arcane names are seriously unintuitive and are a patchwork quilt drawn from a vocabulary from disparate fields over many decades, but they're set in stone now so you just have to find a mnemonic to learn them and not get too hung up on them and get on with things...

$\endgroup$
0
$\begingroup$

Disclaimer: I am not a native speaker either.

The first thing that came to my mind is you hear some times in the news that some automaker needs to recall some vehicles because of some issues. But usually, they only recall cars based on some criteria. So not all cars with the problem are "recalled". To rephrase, while some of those with the issues (true positive) are recalled, some others with the issues may not qualify for the criteria of the recall (false negative).

I think sensitivity can be interpreted from the angle of the "rule/criteria of the recall". A higher sensitivity means the rule is more sensitive to the problem (E.g. the issues of a car), but this will likely result in a higher false positive.

So overall, recall is among the true positives the percentage successfully identified (recalled). Sensitivity is how sensitive the classification rule/algorithm is to the problem (attributes of the true positive).

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.