The nearest neighbor approach a$.$ is computationally independent of dataset size. b$.$ requires a numeric output attribute. c$.$ is limited to classifying datasets with numeric input attributes. d$.$ stores instances rather than a generalized model of the data.