- Explore scientific data with a set of tools for human-guided or automated discovery
- Design & configure data processing pipelines
- Define the parameter ranges for your algorithms, available algorithmic choices, and the framework will generate pipeline instances for you
- Use automatically perturbed data processing pipelines to create different data products.
- Easy to use with scikit-dataaccess for integration of a variety of scientific data sets
Install
pip install scikit-discovery
Documentation
See https://github.com/MITHaystack/scikit-discovery/tree/master/skdiscovery/docs
Contributors
Project lead: Victor Pankratius (MIT)
Contributors: Cody M. Rude, Justin D. Li, David M. Blair, Michael G. Gowanlock, Evan Wojciechowski, Victor Pankratius
Acknowledgements
We acknowledge support from NASA AIST14-NNX15AG84G, NASA AIST16-80NSSC17K0125, NSF ACI-1442997, NSF AGS-1343967, and Amazon AWS computing access support.
Examples
Example code with complete science case studies are available as Jupyter Notebooks at:
