In machine learning terminology, what is “true positive”?

The concept of "true positive" refers specifically to instances in which a model accurately identifies positive cases. In the context of a binary classification problem, this means that when the model predicts a positive outcome, and the actual outcome is indeed positive, it counts as a true positive. This measurement is crucial in evaluating the performance of predictive models, especially in fields like healthcare, fraud detection, or any binary classification system, where it’s critical to understand how many of the predicted positive cases are genuine.

Accurately identifying true positives allows practitioners to gauge the model's effectiveness, particularly with respect to sensitivity or recall, which is the ability of a model to find all relevant cases within the data. In scenarios where identifying the positive class is of primary importance, such as diagnosing diseases, having a high number of true positives can significantly impact decision-making and patient outcomes.