… actually negative, or in our example, individuals the model classifies as terrorists that are not. While recall expresses the ability to find all relevant instances in a dataset, precision expresses the proportion of the data points our model says was relevant actually were relevant.