There are instances when the data collected from a specific population segment is limited, e.g., when only a limited amount of learning data has been collected or when some features of the discriminated population segment are missing. This can lead to the model being inadequate for predicting that population.