Comparison of source count distributions as a function of flux for the estimated number of pulsar and OTHER sources among the unassociated sources in the two- and three-class cases for oversampling-by-repeating and SMOTE. The number of pulsar-like sources in the two-class case is not corrected for the presence of OTHER sources. The changes due to using the SMOTE oversampling technique are comparable to the difference among the algorithms for oversampling-by-repeating.