Fig. 4From: Diverse approaches to predicting drug-induced liver injury using gene-expression profilesPhase I results of hyperparameter optimization based on the “number of trees” hyperparameter for the Random Forests algorithm. We used an ad hoc approach to tune algorithm hyperparameters on the training set. As an example, we tuned the “number of trees” hyperparameter for the Random Forests algorithm. The performance varied considerably for different numbers of trees. All 4 metrics peaked near 25 trees (red lines). MCC = Matthews correlation coefficientBack to article page