Fig. 11

The figure displays the mean F1 scores of our split model across five models from a 5-fold cross-validation, evaluated for each functional group. These scores were calculated using a holdout test set