dislib is a distributed computing library highly focused on machine learning on top of PyCOMPSs. Inspired by NumPy and scikit-learn, dislib provides various supervised and unsupervised learning algorithms through an easy-to-use API.
Contact:
Primary tabs
New Features
- Added grid search and randomized search with cross-validation
- Added K-fold splitter
- dislib command line can now run jupyter notebooks
Bug Fixes
- Fixed various bugs in fancy indexing of ds-arrays
- dislib command line now works on MacOS
- Fixed "source" links in the documentation to point to the appropriate version of the source code
- dislib command line now works even if PyCOMPSs is not installed
Improvements
- Added a new notebook and improved the existing one
- PCA now supports sparse data
- Estimators now extend scikit-learn's base estimator for greater integration