The RAMP ecosystem contains two organizations and two libraries. The purpose of the bundle is to define, build, manage, and optimize data analytics workflows, typically on the top of open source machine learning libraries like pandas, scikit-learn, and keras. The bundle consists of
Library/Organization | Purpose | Publicly available |
---|---|---|
ramp-workflow | A set of reusable tools and scripts to define score types (metrics), workflow elements, prediction types and data connectors. | โ |
ramp-board | A library managing the frontend and the database of the RAMP platform. | ๐ซ |
ramp-data | An organization containing data sets on which workflows are trained and evaluated. | ๐ซ |
ramp-kits | An organization containing starting kits that use tools from ramp-workflow to implement a first valid (tested) workflow. | โ |
- I am a data science teacher
- I am a data science student or novice data scientist
- I am a practicing data scientist
- I am a researcher in machine learning
- I am a researcher in a domain science or I have a predictive problem in my business
- Install the latest
ramp-workflow
library
$ pip install https://api.github.com/repos/paris-saclay-cds/ramp-workflow/zipball/master
This will set up some command line scripts like ramp_test_submission
.
We suggest to use a dedicated virtual environment if you are familiar with it.
- Pick a starting-kit on https://github.com/ramp-kits
Clone it locally and fire up the starting kit notebook.
It will guide you through the problem, describe the data and the workflow, and let you run the pipeline.
Fore more details, visit the wiki.
Contribute to ramp-workflow
ramp-workflow
is meant to be a collaborative library. We value external contributions.
Refer to this wiki page.