All Versions
34
Latest Version
Avg Release Cycle
20 days
Latest Release
552 days ago
Changelog History
Page 4
Changelog History
Page 4
-
v0.14.2 Changes
April 17, 2019๐ New features
- ๐ฆ Add
has_header
parameter toCatboostEvaluation
class.
๐ฅ Breaking changes
- ๐ Change output feature indices separator (
:
to;
) in theCatboostEvaluation
class.
- ๐ฆ Add
-
v0.14.1 Changes
April 13, 2019๐ฅ Breaking changes
- ๐ Changed default value for
--counter-calc-method
option toSkipTest
๐ New features:
- ๐ Add guid to trained models. You can access it in Python using
get_metadata
function, for exampleprint catboost_model.get_metadata()['model_guid']
๐ Bug fixes and other changes:
- Compatibility with glibc 2.12
- ๐ Improved embedded documentation
- ๐ Improved warning and error messages
- ๐ Changed default value for
-
v0.14.0 Changes
April 09, 2019๐ New features:
๐ GPU training now supports several tree learning strategies, selectable with
grow_policy
parameter. Possible values:SymmetricTree
-- The tree is built level by level untilmax_depth
is reached. On each iteration, all leaves from the last tree level will be split with the same condition. The resulting tree structure will always be symmetric.Depthwise
-- The tree is built level by level untilmax_depth
is reached. On each iteration, all non-terminal leaves from the last tree level will be split. Each leaf is split by condition with the best loss improvement.Lossguide
-- The tree is built leaf by leaf untilmax_leaves
limit is reached. On each iteration, non-terminal leaf with best loss improvement will be split.
Note: grow policies
Depthwise
andLossguide
currently support only training and prediction modes. They do not support model analysis (like feature importances and SHAP values) and saving to different model formats like CoreML, ONNX, and JSON.- The new grow policies support several new parameters:
0๏ธโฃmax_leaves
-- Maximum leaf count in the resulting tree, default 31. Used only forLossguide
grow policy. Warning: It is not recommended to set this parameter greater than 64, as this can significantly slow down training.
min_data_in_leaf
-- Minimum number of training samples per leaf, default 1. CatBoost will not search for new splits in leaves with sample count less thanmin_data_in_leaf
. This option is available forLossguide
andDepthwise
grow policies only.
Note: the new types of trees will be at least 10x slower in prediction than default symmetric trees.
๐ GPU training also supports several score functions, that might give your model a boost in quality. Use parameter
score_function
to experiment with them.Now you can use quantization with more than 255 borders and
one_hot_max_size
> 255 in CPU training.๐ New features in Python package:
- It is now possible to use
save_borders()
function to write borders to a file after training. - Functions
predict
,predict_proba
,staged_predict
, andstaged_predict_proba
now support applying a model to a single object, in addition to usual data matrices.
Speedups:
- ๐ Impressive speedups for sparse datsets. Will depend on the dataset, but will be at least 2--3 times for sparse data.
๐ฅ Breaking changes:
- ๐ฆ Python-package class attributes don't raise exceptions now. Attributes return
None
if not initialized. - โก๏ธ Starting from 0.13 we have new feature importances for ranking modes. The new algorithm for feature importances shows how much features contribute to the optimized loss function. They are also signed as opposed to feature importances for not ranking modes which are non negative. This importances are expensive to calculate, thus we decided to not calculate them by default during training starting from 0.14. You need to calculate them after training.
-
v0.13.1 Changes
March 20, 2019๐ Changes:
- ๐ Fixed a bug in shap values that was introduced in v0.13