Fig. 6
From: Unsupervised encoding selection through ensemble pruning for biomedical classification
Overview of the cross validation. (Orange) In the preprocessing the train/test splits are created, and the kappa-error diagram is calculated. (Yellow) The pruning methods, including MVO, select the best encodings and base classifiers based on the validation data. (Blue) All models are trained with the entire training data and tested on the unseen test set. (Purple) The results are collected, statistically validated, and illustrated