ballet 0.5.1

Core functionality for lightweight, collaborative data science projects

ballet

A lightweight framework for collaborative data science projects through feature engineering.

Ballet projects maintain a feature engineering pipeline invariant: at any point, the code and features within a project repository can be used for end-to-end feature engineering for a given dataset. To expand on an existing feature engineering pipeline, well-structured feature source code submissions can be proposed by contributors and extensively validated for compatibility and performance.

Ballet provides the following functionality:

• ballet-quickstart, a command to generate a new predictive modeling project that uses Ballet framework
• Feature objects, that store feature metadata as well as a robust DelegatingRobustTransformer transformer pipeline built alongside the sklearn_pandas project.
• ballet.eng, a library of versatile transformers and transformer building blocks for developing features that learn.
• an extensive feature validation suite, that checks project structure and feature API adherence and runs a streaming logical feature selection algorithm.

Ballet is under active development, please report all bugs.

• Free software: MIT license
• Documentation: https://hdi-project.github.io/ballet
• Homepage: https://github.com/HDI-Project/ballet

History

0.5 (2018-10-14)

• Add project template and ballet-quickstart command
• Add project structure checks and feature API checks
• Implement multi-stage validation routine driver

0.4 (2018-09-21)

• Implement Modeler for versatile modeling and evaluation
• Change project name

0.3 (2018-04-28)

• Implement PullRequestFeatureValidator
• Add util.travis, util.modutil, util.git util modules

0.2

• Implement ArrayLikeEqualityTestingMixin
• Implement get_contrib_features

0.1

• First release on PyPI