Fig. 3.

Confusion matrices for the read-based approach. a Confusion matrix for the random forest model trained on a random 70/30 train/test data partition. b Confusion matrix for the random forest model trained on a random 70/30 train/test data partition of the rare-species-removed data set