Precision as a function of S/N for the three classes. The solid lines correspond to the one-stage classification, and the dashed lines correspond to the two-stage classification. The precision levels increase with increasing S/N, but these trends flatten at S/N ≳ 20 for pre- and post-mergers. The mean errors on the precision (derived from bootstrapping) are similar between the one-stage and two-stage set-ups and are indicated in the top left corner. The horizontal dotted line at around 33% corresponds to the precision of a random classifier.