Fig. 3.

Random Forest model performance. Top row: Training and prediction accuracy across models trained on (A) Lund2017, (B) TCGA and (C) mixed Lund2017/TCGA data Bottom row: Average score difference between the predicted and second closest class