Toolkit for building predictive workflows on the top of pandas and scikit-learn.
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
rampwf
.coveragerc
.gitattributes
.gitignore
.travis.yml
CHANGELOG.md
LICENSE
MANIFEST.in
README.md
appveyor.yml
requirements.txt
setup.cfg
setup.py
testing-requirements.txt
versioneer.py

README.md

The RAMP ecosystem

Build Status Build status Coverage License Docs

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.
ramp-board A library managing the frontend and the database of the RAMP platform. 🚫
ramp-data An organization containing data sets on which workflows are trained and evaluated. 🚫
ramp-kits An organization containing starting kits that use tools from ramp-workflow to implement a first valid (tested) workflow.

Why do I want this bundle ?

Getting started

  1. 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.

  1. 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.