Used for more significant model transformations. Examples: Data needs to be aggregated (e.g. batches). Multiple sources needs to be unified into a single model.
