Recent Releases of stree

stree - v1.4.0: Upgrade requirements

  • Change python minor version to 3.11
  • Change build tool to hatch
  • Update build source from setup.py to pyproject.toml
  • Update github action versions
  • Remove black from github actions linting as it wouldn't read from pyproject.toml
  • Add call method to Stree to support sklearn ensembles requirements for base estimators
  • Update tests
  • Update readthedocs configuration

- Python
Published by rmontanana over 1 year ago

stree - v1.3.2: Fixing minor issues

Working nicely with scikit-learn v 1.4.2 and python 3.11.5

- Python
Published by rmontanana almost 2 years ago

stree - v1.3.1: update to sklearn 1.2

Update to sklearn 1.2

- Python
Published by rmontanana about 3 years ago

stree - v1.3.0: predict_proba

- Python
Published by rmontanana over 3 years ago

stree -

- Python
Published by rmontanana almost 4 years ago

stree -

Fix random seed not used in fsmutual (mutualinfoclassif from sklearn.featureselection) that produced flaky tests in odte

- Python
Published by rmontanana over 4 years ago

stree -

  • Add iwss and true random (only one combination generated) feature selection

  • Update docs and source comments

  • Remove obsolete binder links

- Python
Published by rmontanana over 4 years ago

stree -

Add feature selection to node splits: f-value, mutual information, cfs and fcbf

- Python
Published by rmontanana over 4 years ago

stree -

- Python
Published by rmontanana over 4 years ago

stree -

Added the following features: * libsvm linear kernel * normalization in a per node basis * feature selection based on mutual information w.r.t. the label * feature selection based on f-values ranking * implement one versus one strategy in multi class

- Python
Published by rmontanana almost 5 years ago

stree - Primera versión estable

Versión utilizada en la experimentación referida en el paper "STree: a single multi-class oblique decision tree based on support vector machines."

- Python
Published by rmontanana almost 5 years ago