Comparison between classifications from ZTF DI-Ap vs. ZTF alerts from Sánchez-Sáez et al. (2021) (left) and DR11-psf from SS23 (right). We only include in these comparisons sources with probabilities greater than 0.5 in the final class. We divided each row by the number of objects per class with DI-Ap labels (with P ≥ 0.5), which were included in the analysis, and we rounded these percentages to integer values.