What are true positives?
True positives are the cases a classifier labels as positive that really are positive. When a model predicts that a container shipment will miss its delivery window and the shipment does arrive late, that prediction is a true positive. The count of true positives occupies one cell of the confusion matrix and appears in the numerator of both precision and recall.
Why true positives matter
True positives are where a model earns its keep, because each one is a case where acting on the prediction can change an outcome. A freight forwarder uses a delay-prediction model to warn customers and rebook capacity before a shipment goes late. Only true positives create that value. False positives waste a rebooking, and false negatives leave a customer unwarned.
Counting true positives requires knowing the real outcome, which can arrive days after the prediction. The forwarder reconciles each prediction with the actual delivery timestamp once it is recorded, so the count is available for weekly reporting. A raw true positive count means little without the false positive and false negative counts beside it, which is why the confusion matrix shows all four cells together. The definition of positive must also be fixed in advance. If the team later redefines late as more than four hours rather than more than one, every count changes.