Skip to main content
Fig. 5 | BioData Mining

Fig. 5

From: Unsupervised encoding selection through ensemble pruning for biomedical classification

Fig. 5

Overview of the workflow. (a) For each fold of the MCCV, the preprocessing is conducted, i.e., the indices of the train/test splits are determined, and the data is scaled. (b) The pruning methods, e.g., Pareto frontier and MVO, select the current fold’s encodings and the number of base classifiers. (c) Different ensembles with various base classifiers are trained and validated on the test data. (d) The results are collected, statistically validated, and illustrated. The workflow accepts an arbitrary number of datasets as input (arrows). For each dataset (bold arrows), the steps a to d are executed successively. Refer to the method section for more details

Back to article page