Fig. 3
From: SIMPD: an algorithm for generating simulated time splits for validating machine learning approaches
Validation performance of random forest models on the NIBR medicinal chemistry project data sets. The black dashed line and right hand y-axis show the AUC values for the temporal splits. The gray triangles and blue squares show \(\Delta\) AUC values (left hand y-axis) for the random and neighbor splits, respectively. The points are ordered by decreasing temporal AUC values