Ensembles and PMML in KNIME
Alexander Fillbrunn, Iris Adä, Thomas R. Gabriel, Michael R. Berthold · KOPS (University of Konstanz) · 2013
In this paper we describe how ensembles can be trained, modified and applied in the open source data analysis platform, KNIME. We focus on recent extensions that also allow ensembles, represented in PMML, to be processed. This way ensembles generated in KNIME can be deployed to PMML scoring engines. In addition ensembles created by other tools and represented as PMML can be applied or further processed (modified or filtered) using intuitive KNIME workflows.