A Quick Overview of Evaluation Metrics for Classification Models

Recall is the ability of the classifier to find all the positive samples. In other words, recall is the metric that answer the question: "From all the instances that belongs to positive class, how many the model labeled as positive?" For instance, imagine you want to predict who is going to vote on candidate "A" next election. In a population of 100 persons, 10 indeed voted candidate "A". Supposing the model labeled all 100 persons as positive (voted candidate "A") the recall would be 100% because the model found all the positive cases.