AIExplainer
Machine Learning Intermediate 2 min read

What is recall?

The proportion of relevant items that are correctly identified by a model

Recall is a measure of how well a model can find all the relevant items in a dataset. It's like finding all the needles in a haystack. A high recall means the model is good at finding most of the relevant items, but it might also find some irrelevant ones.

A librarian trying to find all the books on a specific topic. If the librarian finds 9 out of 10 books, but also picks up 2 irrelevant books, their recall is high because they found most of the relevant books, but their precision might be low because they also picked up some irrelevant ones.

A search engine's recall is high if it returns most of the relevant websites for a given search query, even if it also returns some irrelevant ones.

Recall is used to evaluate the performance of models in information retrieval, classification, and other tasks where finding relevant items is important.

Some people think that recall is the same as accuracy, but it's not. Recall only cares about finding relevant items, while accuracy cares about finding both relevant and irrelevant items correctly.

The concept of recall has been around since the early days of information retrieval, and it's still widely used today in many fields, including machine learning and data science.

sensitivity true positive rate detection rate

Three products for different needs — explore what’s relevant to you.