# Idea The terms [[recall]] (and [[precision]]) comes from information retrieval theory and document retrieval systems (predate machine learning). It measures the ability of a system to "recall" all the relevant instances from a dataset: "*Of all the positive instances (that ought to be identified), how many did we correctly or manage to identify?*" (see [[confusion matrix]]). **What proportion of actual positives was identified correctly by the model?** Goal: We care about getting each and every positive sample correctly classified. Recall (also known as **sensitivity** in binary classification) is defined as the following (contrast with [[precision]] and [[accuracy]]): $\frac{true \ positive}{true \ positive + positive \ ones \ that \ were \ not \ retrieved}$ $\frac{true \ positive}{true \ positive + false \ negative}$ $\frac{true \ positive}{total \ actual \ positive}$ See [[confusion matrix]] for another way to visualize recall. It measures **how complete or good the results** or how good the system is at retrieving all relevant documents. It measures how many actual positives the model captures. ![[s20230523_125150 1.png|400]] Mnemonic: **P** stands for "pie" (circle); **rec** stands for rectangle. The denominator in precision is a circle (pie); the denominator in recall is a rectangle. Assuming we're classifying spam, it tells us the proportion of messages that were actually spam were classified by us as spam. Recall is the complement of [[type 2 error]] rate. Search engine query example - search returns 30 pages; 20 are relevant - bufailed to return 40 additional relevant pages - precision: 20/30 = 2/3 ("how valid the results are") - recall: 20/60 = 1/3 ("how complete the results are") ## Significance of recall Use recall when **false negatives** are costlier or detrimental than false positives. For example, in medical diagnostics, missing a positive case (false negative) can lead to death. High recall minimizes this risk. To balance recall and precision, use [[F1 score]] or the [[general F-beta score]]. # References - https://en.wikipedia.org/wiki/Precision_and_recall - [You Will Never Forget Precision and Recall If You Use the Mindset Technique](https://www.blog.dailydoseofds.com/p/you-will-never-forget-precision-and?publication_id=1119889&post_id=140686520&isFreemail=true&r=35ltar)