Sharing rapidminer workows and experiments with OpenML
Jan N. van Rijn, Joaquin Vanschoren · 2015
OpenML is an online, collaborative environment for machine learning where researchers and practitioners can share datasets, workflows and experiments. While it is integrated in several machine learning environments, it was not yet integrated into environments that offer a graphical interface to easily build and experiment with many data analysis workflows. In this work we introduce an integration into the popular RapidMiner environment, that will allow RapidMiner users to import data directly from OpenML and automatically share all their workflows and experiments. OpenML will then link these results to all other results obtained by other people, possibly with other tools, creating a single connected overview of the best workflows on a large set of machine learning problems. This is useful to learn and build on the results of others, to collaborate with many people online, and it provides a wealth of information to study how to construct workflows for new machine learning problems. We demonstrate the capabilities of this integration and identify several research opportunities.