The RAMP framework: from reproducibility to transparency in the design and optimization of scientific workflows

Balázs Kégl, Alexandre Boucaud, Mehdi Cherti, Akin Kazakci, Alexandre Gramfort, Guillaume Lemaitre, Joris Van den Bossche, Djalel Benbouzid, Camille Marini

Abstract: The RAMP (Rapid Analytics and Model Prototyping) is a software and project management tool developed by the Paris-Saclay Center for Data Science. The original goal was to accelerate the adoption of high-quality data science solutions for domain science problems by running rapid collaborative prototyping sessions. Today it is a full-blown data science project management tool promoting reproducibility, fair and transparent model evaluation, and democratization of data science. We have used the framework for setting up and solving about twenty scientific problems, for organizing scientific sub-communities around these events, and for training novice data scientists.

TL;DR: We describe the RAMP framework that we have been using for organizing work around reproducible data science workflows for scientific projects.

Keywords: machine learning, data science, reproducible workflow, data challenge