Confusion matrices from FOF-AM compared to DfL for the same model and data pairings. Here the performance statistics are shown for the full sample of all galaxies, with galaxies in the extended satellite region treated as normal satellites. The overall accuracy of both methods are similar, but DfL achieves notably higher precision in the central galaxy sample, which is important for accurately identifying haloes. On the other hand, the recall in central galaxies is lower in DfL compared to FOF-AM, indicating a trade off between purity and completeness.