Precision TP+FPTP How many samples are actually positive out of the total number of predicted positive samples? → How precise is the model in predicting positive samples?