- Dropped support for Daru
- Dropped support for Ruby < 3.3
- Fixed memory leak with
cvmethod
- Added ARM shared library for Linux
- Improved performance of
predictmethod forpandas_categorical
- Updated LightGBM to 4.6.0
- Added support for different prediction types
- Added support for
pandas_categoricaltopredictmethod - Added support for hashes and Rover data frames to
predictmethod - Added support for hashes to
Dataset - Added
importance_typeoption todump_model,model_to_string, andsave_modelmethods - Changed
Datasetto use column names for feature names with Rover and Daru - Changed
predictmethod to match feature names with Daru - Dropped support for Ruby < 3.1
- Updated LightGBM to 4.5.0
- Updated LightGBM to 4.4.0
- Updated LightGBM to 4.3.0
- Updated LightGBM to 4.1.0
- Updated LightGBM to 4.0.0
- Fixed error with
dupandclone - Dropped support for Ruby < 3
- Updated LightGBM to 3.3.5
- Improved ARM detection
- Updated LightGBM to 3.3.0
- Added
feature_namemethod to boosters
- Updated LightGBM to 3.2.0
- Added ARM shared library for Mac
- Updated LightGBM to 3.1.1
- Updated LightGBM to 3.1.0
- Updated LightGBM to 3.0.0
- Made
best_iterationandeval_histconsistent with Python
- Added support for Rover
- Improved performance of Numo datasets
- Improved error message when OpenMP not found on Mac
- Fixed
Cannot add validation dataerror
- Updated LightGBM to 2.3.1
- Switched to doubles for datasets and predictions
- Updated LightGBM to 2.3.0
- Fixed error with JRuby
- Packaged LightGBM with gem
- Added support for missing values
- Added
feature_namesto datasets - Fixed Daru training and prediction
- Friendlier message when LightGBM not found
- Added
Ranker - Added early stopping to Scikit-Learn API
- Free memory when objects are destroyed
- Removed unreleased
dump_textmethod
- Added Scikit-Learn API
- Added support for Daru and Numo::NArray
- Added
cvmethod - Added early stopping
- Fixed multiclass classification
- Added training API
- Added many methods
- First release