Skip to content

Latest commit

 

History

History
1247 lines (971 loc) · 80.8 KB

RELEASE.md

File metadata and controls

1247 lines (971 loc) · 80.8 KB

Release 0.24.4

Speedup

  • Major speedup asymmetric trees training time on CPU (2x speedup on Epsilon with 16 threads). We would like to recognize Intel software engineering team’s contributions to Catboost project.

New features

  • Now we publish Python 3.9 wheels. Related issues: #1491, #1509, #1510
  • Allow boost_from_average for MultiRMSE loss.
  • Add tag pairwise=False for sklearn compatibility. Fixes issue #1518

Bugfixes:

  • Allow fstr calculation for datasets with embeddings
  • Fix feature_importances_ for fstr with texts
  • Virtual ensebles fix: use proper unshrinkage coefficients
  • Fixed constants in RMSEWithUnceratainty loss function calculation to correspond values from original paper
  • Allow shap values calculation for model with zero-weights and non-zero leaf values. Now we use sum of leaf weights on train and current dataset to guarantee non-zero weights for leafs, reachable on current dataset. Fixes issues #1512, #1284

Release 0.24.3

New functionality

  • Support fstr text features and embeddings. Issue #1293

Bugfixes:

  • Fix model apply speed regression from 0.24.1 & 0.24.2
  • Different fixes in embeddings support: fixed apply and model serialization, fixed apply on texts and embeddings
  • Fixed virtual ensembles prediction - use proper scaling, fix apply (issue #1462)
  • Fix score() method for RMSEWithUncertainty issue #1482
  • Automatically use correct prediction_type in score()

Release 0.24.2

Uncertainty prediction

  • Supported uncertainty prediction for classification models.
  • Fixed RMSEWithUncertainty data uncertainty prediction - now it predicts variance, not standard deviation.

New functionality

  • Allow categorical feature counters for MultiRMSE loss function.
  • group_weight parameter added to catboost.utils.eval_metric method to allow passing weights for object groups. Allows correctly match weighted ranking metrics computation when group weights present.
  • Faster non-owning deserialization from memory with less memory overhead - moved some dynamically computed data to model file, other data is computed in lazy manner only when needed.

Experimental functionality

  • Supported embedding features as input and linear discriminant analysis for embeddings preprocessing. Try adding your embeddings as new columns with embedding values array in Pandas.Dataframe and passing corresponding column names to Pool constructor or fit function with embedding_features=['EmbeddingFeaturesColumnName1, ...] parameter. Another way of adding your embedding vectors is new type of column in Column Description file NumVector and adding semicolon separated embeddings column to your XSV file: ClassLabel\t0.1;0.2;0.3\t....

Educational materials

  • Published new tutorial on uncertainty prediction.

Bugfixes:

  • Reduced GPU memory usage in multi gpu training when there is no need to compute categorical feature counters.
  • Now CatBoost allows to specify use_weights for metrics when auto_class_weights parameter is set.
  • Correctly handle NaN values in plot_predictions function.
  • Fixed floating point precision drop releated bugs during Multiclass training with lots of objects in our case, bug was triggered while training on 25mln objects on single GPU card.
  • Now average parameter is passed to TotalF1 metric while training on GPU.
  • Added class labels checks
  • Disallow feature remapping in model predict when there is empty feature names in model.

Release 0.24.1

Uncertainty prediction

Main feature of this release is total uncertainty prediction support via virtual ensembles. You can read the theoretical background in the preprint Uncertainty in Gradient Boosting via Ensembles from our research team. We introduced new training parameter posterior_sampling, that allows to estimate total uncertainty. Setting posterior_sampling=True implies enabling Langevin boosting, setting model_shrink_rate to 1/(2*N) and setting diffusion_temperature to N, where N is dataset size. CatBoost object method virtual_ensembles_predict splits model into virtual_ensembles_count submodels. Calling model.virtual_ensembles_predict(.., prediction_type='TotalUncertainty') returns mean prediction, variance (and knowledge uncertrainty for models, trained with RMSEWithUncertainty loss function). Calling model.virtual_ensembles_predict(.., prediction_type='VirtEnsembles') returns virtual_ensembles_count predictions of virtual submodels for each object.

New functionality

  • Supported non-owning model deserialization for models with categorical feature counters

Speedups

  • We've done lot's of speedups for sparse data loading. For example, on bosch sparse dataset preprocessing speed got 4.5x speedup while running in 28 thread setting.

Bugfixes:

  • Fixed target check for PairLogitPairwise on GPU. Issue #1217
  • Supported n_features_in_ attribute required for using CatBoost in sklearn pipelines. Issue #1363

Release 0.24

New functionality

  • We've finally implemented MVS sampling for GPU training. Switched default bootstrap algorithm to MVS for RMSE loss function while training on GPU
  • Implemented near-zero cost model deserialization from memory blob. Currently, if your model doesn't use categorical features CTR counters and text features you can deserialize model from, for example, memory-mapped file.
  • Added ability to load trained models from binary string or file-like stream. To load model from bytes string use load_model(blob=b'....'), to deserialize form file-like stream use load_model(stream=gzip.open('model.cbm.gz', 'rb'))
  • Fixed auto-learning rate estimation params for GPU
  • Supported beta parameter for QuerySoftMax function on CPU and GPU

New losses and metrics

  • New loss function RMSEWithUncertainty - it allows to estimate data uncertainty for trained regression models. The trained model will give you a two-element vector for each object with the first element as regression model prediction and the second element as an estimation of data uncertainty for that prediction.

Speedups

  • Major speedups for CPU training: kdd98 -9%, higgs -18%, msrank -28%. We would like to recognize Intel software engineering team’s contributions to Catboost project. This was mutually beneficial activity, and we look forward to continuing joint cooperation.

Bugfixes:

  • Fixed CatBoost model export as Python code
  • Fixed AUC metric creation
  • Add text features to model.feature_names_. Issue#1314
  • Allow models, trained on datasets with NaN values (Min treatment) and without NaNs in model_sum() or as the base model in init_model=. Issue #1271

Educational materials

  • Published new tutorial on categorical features parameters. Thanks @garkavem

Release 0.23.2

New functionality

  • Added plot_partial_dependence method in python-package (Now it works for models with symmetric trees trained on dataset with numerical features only). Implemented by @felixandrer.
  • Allowed using boost_from_average option together with model_shrink_rate option. In this case shrinkage is applied to the starting value..
  • Added new auto_class_weights option in python-package, R-package and cli with possible values Balanced and SqrtBalanced. For Balanced every class is weighted maxSumWeightInClass / sumWeightInClass, where sumWeightInClass is sum of weights of all samples in this class. If no weights are present then sample weight is 1. And maxSumWeightInClass - is maximum sum weight among all classes. For SqrtBalanced the formula is sqrt(maxSumWeightInClass / sumWeightInClass). This option supported in binclass and multiclass tasks. Implemented by @egiby.
  • Supported model_size_reg option on GPU. Set to 0.5 by default (same as in CPU). This regularization works slightly differently on GPU: feature combinations are regularized more aggressively than on CPU. For CPU cost of a combination is equal to number of different feature values in this combinations that are present in training dataset. On GPU cost of a combination is equal to number of all possible different values of this combination. For example, if combination contains two categorical features c1 and c2, then the cost will be #categories in c1 * #categories in c2, even though many of the values from this combination might not be present in the dataset.
  • Added calculation of Shapley values, (see formula (2) from https://arxiv.org/pdf/1802.03888.pdf). By default estimation from this paper (Algorithm 2) is calcucated, that is much more faster. To use this mode specify shap_calc_type parameter of CatBoost.get_feature_importance function as "Exact". Implemented by @LordProtoss.

Bugfixes:

  • Fixed onnx converter for old onnx versions.

Release 0.23.1

New functionality

  • CatBoost model could be simply converted into ONNX object in Python with catboost.utils.convert_to_onnx_object method. Implemented by @monkey0head
  • We now print metric options with metric names as metric description in error logs by default. This allows you to distinguish between metrics of the same type with different parameters. For example, if user sets weigheted average TotalF1 metric CatBoost will print TotalF1:average=Weighted as corresponding metric column header in error logs. Implemented by @ivanychev
  • Implemented PRAUC metric (issue #737). Thanks @azikmsu
  • It's now possible to write custom multiregression objective in Python. Thanks @azikmsu
  • Supported nonsymmetric models export to PMML
  • class_weights parameter accepts dictionary with class name to class weight mapping
  • Added _get_tags() method for compatibility with sklearn (issue #1282). Implemented by @crazyleg
  • Lot's of improvements in .Net CatBoost library: implemented IDisposable interface, splitted ML.NET compatible and basic prediction classes in separate libraries, added base UNIX compatibility, supported GPU model evaluation, fixed tests. Thanks @khanova
  • In addition to first_feature_use_penalties presented in the previous release, we added new option per_object_feature_penalties which considers feature usage on each object individually. For more details refer the tutorial.

Breaking changes

  • From now on we require explicit loss_function param in python cv method.

Bugfixes:

  • Fixed deprecation warning on import (issue #1269)
  • Fixed saved models logging_level/verbose parameters conflict (issue #696)
  • Fixed kappa metric - in some cases there were integer overflow, switched accumulation types to double
  • Fixed per float feature quantization settings defaults

Educational materials

  • Extended shap values tutorial with summary plot examples. Thanks @azanovivan02

Release 0.23

New functionality

  • It is possible now to train models on huge datasets that do not fit into CPU RAM. This can be accomplished by storing only quantized data in memory (it is many times smaller). Use catboost.utils.quantize function to create quantized Pool this way. See usage example in the issue #1116. Implemented by @noxwell.
  • Python Pool class now has save_quantization_borders method that allows to save resulting borders into a file and use it for quantization of other datasets. Quantization can be a bottleneck of training, especially on GPU. Doing quantization once for several trainings can significantly reduce running time. It is recommended for large dataset to perform quantization first, save quantization borders, use them to quantize validation dataset, and then use quantized training and validation datasets for further training. Use saved borders when quantizing other Pools by specifying input_borders parameter of the quantize method. Implemented by @noxwell.
  • Text features are supported on CPU
  • It is now possible to set border_count > 255 for GPU training. This might be useful if you have a "golden feature", see docs.
  • Feature weights are implemented. Specify weights for specific features by index or name like feature_weights="FeatureName1:1.5,FeatureName2:0.5". Scores for splits with this features will be multiplied by corresponding weights. Implemented by @Taube03.
  • Feature penalties can be used for cost efficient gradient boosting. Penalties are specified in a similar fashion to feature weights, using parameter first_feature_use_penalties. This parameter penalized the first usage of a feature. This should be used in case if the calculation of the feature is costly. The penalty value (or the cost of using a feature) is subtracted from scores of the splits of this feature if feature has not been used in the model. After the feature has been used once, it is considered free to proceed using this feature, so no substruction is done. There is also a common multiplier for all first_feature_use_penalties, it can be specified by penalties_coefficient parameter. Implemented by @Taube03 (issue #1155)
  • recordCount attribute is added to PMML models (issue #1026).

New losses and metrics

  • New ranking objective 'StochasticRank', details in paper.
  • Tweedie loss is supported now. It can be a good solution for right-skewed target with many zero values, see tutorial. When using CatBoostRegressor.predict function, default prediction_type for this loss will be equal to Exponent. Implemented by @ilya-pchelintsev (issue #577)
  • Classification metrics now support a new parameter proba_border. With this parameter you can set decision boundary for treating prediction as negative or positive. Implemented by @ivanychev.
  • Metric TotalF1 supports a new parameter average with possible value weighted, micro, macro. Implemented by @ilya-pchelintsev.
  • It is possible now to specify a custom multi-label metric in python. Note that it is only possible to calculate this metric and use it as eval_metric. It is not possible to used it as an optimization objective. To write a multi-label metric, you need to define a python class which inherits from MultiLabelCustomMetric class. Implemented by @azikmsu.

Improvements of grid and randomized search

  • class_weights parameter is now supported in grid/randomized search. Implemented by @vazgenk.
  • Invalid option configurations are automatically skipped during grid/randomized search. Implemented by @borzunov.
  • get_best_score returns train/validation best score after grid/randomized search (in case of refit=False). Implemented by @rednevaler.

Improvements of model analysis tools

  • Computation of SHAP interaction values for CatBoost models. You can pass type=EFstrType.ShapInteractionValues to CatBoost.get_feature_importance to get a matrix of SHAP values for every prediction. By default, SHAP interaction values are calculated for all features. You may specify features of interest using the interaction_indices argument. Implemented by @IvanKozlov98.
  • SHAP values can be calculated approximately now which is much faster than default mode. To use this mode specify shap_calc_type parameter of CatBoost.get_feature_importance function as "Approximate". Implemented by @LordProtoss (issue #1146).
  • PredictionDiff model analysis method can now be used with models that contain non symmetric trees. Implemented by @felixandrer.

New educational materials

  • A tutorial on tweedie regression
  • A tutorial on poisson regression
  • A detailed tutorial on different types of AUC metric, which explains how different types of AUC can be used for binary classification, multiclassification and ranking tasks.

Breaking changes

  • When using CatBoostRegressor.predict function for models trained with Poisson loss, default prediction_type will be equal to Exponent (issue #1184). Implemented by @garkavem.

This release also contains bug fixes and performance improvements, including a major speedup for sparse data on GPU.

Release 0.22

New features:

  • The main feature of the release is the support of non symmetric trees for training on CPU. Using non symmetric trees might be useful if one-hot encoding is present, or data has little noise. To try non symmetric trees change grow_policy parameter. Starting from this release non symmetric trees are supported for both CPU and GPU training.
  • The next big feature improves catboost text features support. Now tokenization is done during training, you don't have to do lowercasing, digit extraction and other tokenization on your own, catboost does it for you.
  • Auto learning-rate is now supported in CPU MultiClass mode.
  • CatBoost class supports to_regressor and to_classifier methods.

The release also contains a list of bug fixes.

Release 0.21

New features:

  • The main feature of this release is the Stochastic Gradient Langevin Boosting (SGLB) mode that can improve quality of your models with non-convex loss functions. To use it specify langevin option and tune diffusion_temperature and model_shrink_rate. See the corresponding paper for details.

Improvements:

  • Automatic learning rate is applied by default not only for Logloss objective, but also for RMSE (on CPU and GPU) and MultiClass (on GPU).
  • Class labels type information is stored in the model. Now estimators in python package return values of proper type in classes_ attribute and for prediction functions with prediction_type=Class. #305, #999, #1017. Note: Class labels loaded from datasets in CatBoost dsv format always have string type now.

Bug fixes:

  • Fixed huge memory consumption for text features. #1107
  • Fixed crash on GPU on big datasets with groups (hundred million+ groups).
  • Fixed class labels consistency check and merging in model sums (now class names in binary classification are properly checked and added to the result as well)
  • Fix for confusion matrix (PR #1152), thanks to @dmsivkov.
  • Fixed shap values calculation when boost_from_average=True. #1125
  • Fixed use-after-free in fstr PredictionValuesChange with specified dataset
  • Target border and class weights are now taken from model when necessary for feature strength, metrics evaluation, roc_curve, object importances and calc_feature_statistics calculations.
  • Fixed that L2 regularization was not applied for non symmetric trees for binary classification on GPU.
  • [R-package] Fixed the bug that catboost.get_feature_importance did not work after model is loaded #1064
  • [R-package] Fixed the bug that catboost.train did not work when called with the single dataset parameter. #1162
  • Fixed L2 score calculation on CPU

##Other:

  • Starting from this release Java applier is released simultaneously with other components and has the same version.

##Compatibility:

  • Models trained with this release require applier from this release or later to work correctly.

Release 0.20.2

New features:

  • String class labels are now supported for binary classification
  • [CLI only] Timestamp column for the datasets can be provided in separate files.
  • [CLI only] Timesplit feature evaluation.
  • Process groups of any size in block processing.

Bug fixes:

  • classes_count and class_weight params can be now used with user-defined loss functions. #1119
  • Form correct metric descriptions on GPU if use_weights gets value by default. #1106
  • Correct model.classes_ attribute for binary classification (proper labels instead of always 0 and 1). #984
  • Fix model.classes_ attribute when classes_count parameter was specified.
  • Proper error message when categorical features specified for MultiRMSE training. #1112
  • Block processing: It is valid for all groups in a single block to have weights equal to 0
  • fix empty asymmetric tree index calculation. #1104

Release 0.20.1

New features:

  • Have leaf_estimation_method=Exact the default for MAPE loss
  • Add CatBoostClassifier.predict_log_proba(), PR #1095

Bug fixes:

  • Fix usability of read-only numpy arrays, #1101
  • Fix python3 compatibility for get_feature_importance, PR #1090
  • Fix loading model from snapshot for boost_from_average mode

Release 0.20

New submodule for text processing! It contains two classes to help you make text features ready for training:

  • Tokenizer -- use this class to split text into tokens (automatic lowercase and punctuation removal)
  • Dictionary -- with this class you create a dictionary which maps tokens to numeric identifiers. You then use these identifiers as new features.

New features:

  • Enabled boost_from_average for MAPE loss function

Bug fixes:

  • Fixed Pool creation from pandas.DataFrame with discontinuous columns, #1079
  • Fixed standalone_evaluator, PR #1083

Speedups:

  • Huge speedup of preprocessing in python-package for datasets with many samples (>10 mln)

Release 0.19.1

New features:

  • With this release we support Text features for classification on GPU. To specify text columns use text_features parameter. Achieve better quality by using text information of your dataset. See more in Learning CatBoost with text features
  • MultiRMSE loss function is now available on CPU. Labels for the multi regression mode should be specified in separate Label columns
  • MonoForest framework for model analysis, based on our NeurIPS 2019 paper. Learn more in MonoForest tutorial
  • boost_from_average is now True by default for Quantile and MAE loss functions, which improves the resulting quality

Speedups:

  • Huge reduction of preprocessing time for datasets loaded from files and for datasets with many samples (> 10 million), which was a bottleneck for GPU training
  • 3x speedup for small datasets

Release 0.18.1

New features:

  • Now datasets.msrank() returns full msrank dataset. Previously, it returned the first 10k samples. We have added msrank_10k() dataset implementing the past behaviour.

Bug fixes:

  • get_object_importance() now respects parameter top_size, #1045 by @ibuda

Release 0.18

  • The main feature of the release is huge speedup on small datasets. We now use MVS sampling for CPU regression and binary classification training by default, together with Plain boosting scheme for both small and large datasets. This change not only gives the huge speedup but also provides quality improvement!
  • The boost_from_average parameter is available in CatBoostClassifier and CatBoostRegressor
  • We have added new formats for describing monotonic constraints. For example, "(1,0,0,-1)" or "0:1,3:-1" or "FeatureName0:1,FeatureName3:-1" are all valid specifications. With Python and params-file json, lists and dictionaries can also be used

Bugs fixed:

  • Error in Multiclass classifier training, #1040
  • Unhandled exception when saving quantized pool, #1021
  • Python 3.7: RuntimeError raised in StagedPredictIterator, #848

Release 0.17.5

Bugs fixed:

  • System of linear equations is not positive definite when training MultiClass on Windows, #1022

Release 0.17.4

Improvements:

  • Massive 2x speedup for MultiClass with many classes
  • Updated MVS implementation. See Minimal Variance Sampling in Stochastic Gradient Boosting by Bulat Ibragimov and Gleb Gusev at NeurIPS 2019
  • Added sum_models in R-package, #1007

Bugs fixed:

  • Multi model initialization in python, #995
  • Mishandling of 255 borders in training on GPU, #1010

Release 0.17.3

Improvements:

  • New visualization for parameter tuning. Use plot=True parameter in grid_search and randomized_search methods to show plots in jupyter notebook
  • Switched to jemalloc allocator instead of LFalloc in CLI and model interfaces to fix some problems on Windows 7 machines, #881
  • Calculation of binary class AUC is faster up to 1.3x
  • Added tutorial on using fast CatBoost applier with LightGBM models

Bugs fixed:

  • Shap values for MultiClass objective don't give constant 0 value for the last class in case of GPU training. Shap values for MultiClass objective are now calculated in the following way. First, predictions are normalized so that the average of all predictions is zero in each tree. The normalized predictions produce the same probabilities as the non-normalized ones. Then the shap values are calculated for every class separately. Note that since the shap values are calculated on the normalized predictions, their sum for every class is equal to the normalized prediction
  • Fixed bug in rangking tutorial, #955
  • Allow string value for per_float_feature_quantization parameter, #996

Release 0.17.2

Improvements:

  • For metric MAE on CPU default value of leaf-estimation-method is now Exact
  • Speed up LossFunctionChange feature strength computation

Bugs fixed:

  • Broken label converter in grid search for multiclassification, #993
  • Incorrect prediction with monotonic constraint, #994
  • Invalid value of eval_metric in output of get_all_params(), #940
  • Train AUC is not computed because hint skip_train~false is ignored, #970

Release 0.17.1

Bugs fixed:

  • Incorrect estimation of total RAM size on Windows and Mac OS, #989
  • Failure when dataset is a numpy.ndarray with order='F'
  • Disable boost_from_average when baseline is specified

Improvements:

  • Polymorphic raw features storage (2x---25x faster data preparation for numeric features in non-float32 columns as either pandas.DataFrame or numpy.ndarray with order='F').
  • Support AUC metric for CrossEntropy loss on CPU
  • Added datasets.rotten_tomatoes(), a textual dataset
  • Usability of monotone_constraints, #950

Speedups:

  • Optimized computation of CrossEntropy metric on CPUs with SSE3

Release 0.17

New features:

  • Sparse data support
  • We've implemented and set to default boost_from_average in RMSE mode. It gives a boost in quality especially for a small number of iterations.

Improvements:

  • Quantile regression on CPU
  • default parameters for Poisson regression

Speedups:

  • A number of speedups for training on CPU
  • Huge speedups for loading datasets with categorical features represented as pandas.Categorical. Hint: use pandas.Categorical instead of object to speed up loading up to 200x.

Release 0.16.5

Breaking changes:

  • All metrics except for AUC metric now use weights by default.

New features:

  • Added boost_from_average parameter for RMSE training on CPU which might give a boost in quality.
  • Added conversion from ONNX to CatBoost. Now you can convert XGBoost or LightGBM model to ONNX, then convert it to CatBoost and use our fast applier. Use model.load_model(model_path, format="onnx") for that.

Speed ups:

  • Training is ~15% faster for datasets with categorical features.

Bug fixes:

  • R language: get_features_importance with ShapValues for MultiClass, #868
  • NormalizedGini was not calculated, #962
  • Bug in leaf calculation which could result in slightly worse quality if you use weights in binary classification mode
  • Fixed __builtins__ import in Python3 in PR #957, thanks to @AbhinavanT

Release 0.16.4

Bug fixes:

  • Versions 0.16.* had a bug in python applier with categorical features for applying on more than 128 documents.

New features:

  • It is now possible to use pairwise modes for datasets without groups

Improvements:

  • 1.8x Evaluation speed on asymmetrical trees

Release 0.16.3

Breaking changes:

  • Renamed column Feature Index to Feature Id in prettified output of python method get_feature_importance(), because it supports feature names now
  • Renamed option per_float_feature_binarization (--per-float-feature-binarization) to per_float_feature_quantization (--per-float-feature-quantization)
  • Removed parameter inverted from python cv method. Added type parameter instead, which can be set to Inverted
  • Method get_features() now works only for datasets without categorical features

New features

  • A new multiclass version of AUC metric, called AUC Mu, which was proposed by Ross S. Kleiman on NeurIPS 2019, link
  • Added time series cv
  • Added MeanWeightedTarget in fstat
  • Added utils.get_confusion_matrix()
  • Now feature importance can be calculated for non-symmetric trees

Release 0.16.2

Breaking changes:

  • Removed get_group_id() and get_features() methods of Pool class

New model analysis tools:

  • Added PredictionDiff type of get_feature_importance() method, which is a new method for model analysis. The method shows how the features influenced the fact that among two samples one has a higher prediction. It allows to debug ranking models: you find a pair of samples ranked incorrectly and you look at what features have caused that.
  • Added plot_predictions() method

New features:

  • model.set_feature_names() method in Python
  • Added stratified split to parameter search methods
  • Support catboost.load_model() from CPU snapshots for numerical-only datasets
  • CatBoostClassifier.score() now supports y as DataFrame
  • Added sampling_frequency, per_float_feature_binarization, monotone_constraints parameters to CatBoostClassifier and CatBoostRegresssor

Speedups:

  • 2x speedup of multi-classification mode

Bugfixes:

  • Fixed score() for multiclassification, #924
  • Fixed get_all_params() function, #926

Other improvements:

  • Clear error messages when a model cannot be saved

Release 0.16.1

Breaking changes:

  • parameter fold_count is now called cv in grid_search() and randomized_search
  • cv results are now returned from grid_search() and randomized_search() in res['cv_results'] field

New features:

  • R-language function catboost.save_model() now supports PMML, ONNX and other formats
  • Parameter monotone_constraints in python API allows specifying numerical features that the prediction shall depend on monotonically

Bug fixes:

  • Fixed eval_metric calculation for training with weights (in release 0.16 evaluation of a metric that was equal to an optimized loss did not use weights by default, so overfitting detector worked incorrectly)

Improvements:

  • Added option verbose to grid_search() and randomized_search()
  • Added tutorial on grid_search() and randomized_search()

Release 0.16

Breaking changes:

  • MultiClass loss has now the same sign as Logloss. It had the other sign before and was maximized, now it is minimized.
  • CatBoostRegressor.score now returns the value of $R^2$ metric instead of RMSE to be more consistent with the behavior of scikit-learn regressors.
  • Changed metric parameter use_weights default value to false (except for ranking metrics)

New features:

  • It is now possible to apply model on GPU
  • We have published two new realworld datasets with monotonic constraints, catboost.datasets.monotonic1() and catboost.datasets.monotonic2(). Before that there was only california_housing dataset in open-source with monotonic constraints. Now you can use these two to benchmark algorithms with monotonic constraints.
  • We've added several new metrics to catboost, including DCG, FairLoss, HammingLoss, NormalizedGini and FilteredNDCG
  • Introduced efficient GridSearch and RandomSearch implementations.
  • get_all_params() Python function returns the values of all training parameters, both user-defined and default.
  • Added more synonyms for training parameters to be more compatible with other GBDT libraries.

Speedups:

  • AUC metric is computationally very expensive. We've implemented parallelized calculation of this metric, now it can be calculated on every iteration (or every k-th iteration) about 4x faster.

Educational materials:

  • We've improved our command-line tutorial, now it has examples of files and more information.

Fixes:

  • Automatic Logloss or MultiClass loss function deduction for CatBoostClassifier.fit now also works if the training dataset is specified as Pool or filename string.
  • And some other fixes

Release 0.15.2

Breaking changes:

  • Function get_feature_statistics is replaced by calc_feature_statistics
  • Scoring function Correlation is renamed to Cosine
  • Parameter efb_max_conflict_fraction is renamed to sparse_features_conflict_fraction

New features:

  • Models can be saved in PMML format now.

Note: PMML does not have full categorical features support, so to have the model in PMML format for datasets with categorical features you need to use set one_hot_max_size parameter to some large value, so that all categorical features are one-hot encoded

  • Feature names can be used to specify ignored features

Bug fixes, including:

  • Fixed restarting of CV on GPU for datasets without categorical features
  • Fixed learning continuation errors with changed dataset (PR #879) and with model loaded from file (#884)
  • Fixed NativeLib for JDK 9+ (PR #857)

Release 0.15.1

Bug fixes

  • restored parameter fstr_type in Python and R interfaces

Release 0.15

Breaking changes

  • cv is now stratified by default for Logloss, MultiClass and MultiClassOneVsAll.
  • We have removed border parameter of Logloss metric. You need to use target_border as a separate training parameter now.
  • CatBoostClassifier now runs MultiClass if more than 2 different values are present in training dataset labels.
  • model.best_score_["validation_0"] is replaced with model.best_score_["validation"] if a single validation dataset is present.
  • get_object_importance function parameter ostr_type is renamed to type in Python and R.

Model analysis

  • Tree visualisation by @karina-usmanova.
  • New feature analysis: plotting information about how a feature was used in the model by @alexrogozin12.
  • Added plot parameter to get_roc_curve, get_fpr_curve and get_fnr_curve functions from catboost.utils.
  • Supported prettified format for all types of feature importances.

New ways of doing predictions

  • Rust applier by @shuternay.
  • DotNet applier by @17minutes.
  • One-hot encoding for categorical features in CatBoost CoreML model by Kseniya Valchuk and Ekaterina Pogodina.

New objectives

Speedups

  • Speed up of shap values calculation for single object or for small number of objects by @Lokutrus.
  • Cheap preprocessing and no fighting of overfitting if there is little amount of iterations (since you will not overfit anyway).

New functionality

  • Prediction of leaf indices.

New educational materials

  • Rust tutorial by @shuternay.
  • C# tutorial.
  • Leaf indices.
  • Tree visualisation tutorial by @karina-usmanova.
  • Google Colab tutorial for regression in catboost by @col14m.

And a set of fixes for your issues.

Release 0.14.2

New features

Breaking changes

  • Change output feature indices separator (: to ;) in the CatboostEvaluation class.

Release 0.14.1

Breaking changes

  • Changed default value for --counter-calc-method option to SkipTest

New features:

  • Add guid to trained models. You can access it in Python using get_metadata function, for example print catboost_model.get_metadata()['model_guid']

Bug fixes and other changes:

  • Compatibility with glibc 2.12
  • Improved embedded documentation
  • Improved warning and error messages

Release 0.14.0

New features:

  • GPU training now supports several tree learning strategies, selectable with grow_policy parameter. Possible values:

    • SymmetricTree -- The tree is built level by level until max_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 until max_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 until max_leaves limit is reached. On each iteration, non-terminal leaf with best loss improvement will be split.

    Note: grow policies Depthwise and Lossguide 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: max_leaves -- Maximum leaf count in the resulting tree, default 31. Used only for Lossguide 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 than min_data_in_leaf. This option is available for Lossguide and Depthwise grow policies only.

    Note: the new types of trees will be at least 10x slower in prediction than default symmetric trees.

  • 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.

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, and staged_predict_proba now support applying a model to a single object, in addition to usual data matrices.

Speedups:

  • Impressive speedups for sparse datsets. Will depend on the dataset, but will be at least 2--3 times for sparse data.

Breaking changes:

  • Python-package class attributes don't raise exceptions now. Attributes return None if not initialized.
  • 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.

Release 0.13.1

Changes:

  • Fixed a bug in shap values that was introduced in v0.13

Release 0.13

Speedups:

  • Impressive speedup of CPU training for datasets with predominantly binary features (up to 5-6x).
  • Speedup prediction and shap values array casting on large pools (issue #684).

New features:

  • We've introduced a new type of feature importances - LossFunctionChange. This type of feature importances works well in all the modes, but is especially good for ranking. It is more expensive to calculate, thus we have not made it default. But you can look at it by selecting the type of feature importance.
  • Now we support online statistics for categorical features in QuerySoftMax mode on GPU.
  • We now support feature names in cat_features, PR #679 by @infected-mushroom - thanks a lot @infected-mushroom!
  • We've intoduced new sampling_type MVS, which speeds up CPU training if you use it.
  • Added classes_ attribute in python.
  • Added support for input/output borders files in python package. Thank you @necnec for your PR #656!
  • One more new option for working with categorical features is ctr_target_border_count. This option can be used if your initial target values are not binary and you do regression or ranking. It is equal to 1 by default, but you can try increasing it.
  • Added new option sampling_unit that allows to switch sampling from individual objects to entire groups.
  • More strings are interpreted as missing values for numerical features (mostly similar to pandas' read_csv).
  • Allow skip_train property for loss functions in cv method. Contributed by GitHub user @RakitinDen, PR #662, many thanks.
  • We've improved classification mode on CPU, there will be less cases when the training diverges. You can also try to experiment with new leaf_estimation_backtracking parameter.
  • Added new compare method for visualization, PR #652. Thanks @Drakon5999 for your contribution!
  • Implemented __eq__ method for CatBoost* python classes (PR #654). Thanks @daskol for your contribution!
  • It is now possible to output evaluation results directly to stdout or stderr in command-line CatBoost in calc mode by specifying stream://stdout or stream://stderr in --output-path parameter argument. (PR #646). Thanks @towelenee for your contribution!
  • New loss function - Huber. Can be used as both an objective and a metric for regression. (PR #649). Thanks @atsky for your contribution!

Changes:

  • Changed defaults for one_hot_max_size training parameter for groupwise loss function training.
  • SampleId is the new main name for former DocId column in input data format (DocId is still supported for compatibility). Contributed by GitHub user @daskol, PR #655, many thanks.
  • Improved CLI interface for cross-validation: replaced -X/-Y options with --cv, PR #644. Thanks @tswr for your pr!
  • eval_metrics : eval_period is now clipped by total number of trees in the specified interval. PR #653. Thanks @AntPon for your contribution!

R package:

  • Thanks to @ws171913 we made necessary changes to prepare catboost for CRAN integration, PR #715. This is in progress now.
  • R interface for cross-validation contributed by GitHub user @brsoyanvn, PR #561 -- many thanks @brsoyanvn!

Educational materials:

We have also done a list of fixes and data check improvements. Thanks @brazhenko, @Danyago98, @infected-mushroom for your contributions.

Release 0.12.2

Changes:

  • Fixed loading of epsilon dataset into memory
  • Fixed multiclass learning on GPU for >255 classes
  • Improved error handling
  • Some other minor fixes

Release 0.12.1.1

Changes:

  • Fixed Python compatibility issue in dataset downloading
  • Added sampling_type parameter for YetiRankPairwise loss

Release 0.12.1

Changes:

  • Support saving models in ONNX format (only for models without categorical features).
  • Added new dataset to our catboost.datasets() -- dataset epsilon, a large dense dataset for binary classification.
  • Speedup of Python cv on GPU.
  • Fixed creation of Pool from pandas.DataFrame with pandas.Categorical columns.

Release 0.12.0

Breaking changes:

  • Class weights are now taken into account by eval_metrics(), get_feature_importance(), and get_object_importance(). In previous versions the weights were ignored.
  • Parameter random-strength for pairwise training (PairLogitPairwise, QueryCrossEntropy, YetiRankPairwise) is not supported anymore.
  • Simultaneous use of MultiClass and MultiClassOneVsAll metrics is now deprecated.

New functionality:

  • cv method is now supported on GPU.
  • String labels for classes are supported in Python. In multiclassification the string class names are inferred from the data. In binary classification for using string labels you should employ class_names parameter and specify which class is negative (0) and which is positive (1). You can also use class_names in multiclassification mode to pass all possible class names to the fit function.
  • Borders can now be saved and reused. To save the feature quantization information obtained during training data preprocessing into a text file use cli option --output-borders-file. To use the borders for training use cli option --input-borders-file. This functionanlity is now supported on CPU and GPU (it was GPU-only in previous versions). File format for the borders is described here.
  • CLI option --eval-file is now supported on GPU.

Quality improvement:

  • Some cases in binary classification are fixed where training could diverge

Optimizations:

  • A great speedup of the Python applier (10x)
  • Reduced memory consumption in Python cv function (times fold count)

Benchmarks and tutorials:

  • Added speed benchmarks for CPU and GPU on a variety of different datasets.
  • Added benchmarks of different ranking modes. In this tutorial we compare different ranking modes in CatBoost, XGBoost and LightGBM.
  • Added tutorial for applying model in Java.
  • Added benchmarks of SHAP values calculation for CatBoost, XGBoost and LightGBM. The benchmarks also contain explanation of complexity of this calculation in all the libraries.

We also made a list of stability improvements and stricter checks of input data and parameters.

And we are so grateful to our community members @canorbal and @neer201 for their contribution in this release. Thank you.

Release 0.11.2

Changes:

  • Pure GPU implementation of NDCG metric
  • Enabled LQ loss function
  • Fixed NDCG metric on CPU
  • Added model_sum mode to command line interface
  • Added SHAP values benchmark (#566)
  • Fixed random_strength for Plain boosting (#448)
  • Enabled passing a test pool to caret training (#544)
  • Fixed a bug in exporting the model as python code (#556)
  • Fixed label mapper for multiclassification custom labels (#523)
  • Fixed hash type of categorical features (#558)
  • Fixed handling of cross-validation fold count options in python package (#568)

Release 0.11.1

Changes:

  • Accelerated formula evaluation by ~15%
  • Improved model application interface
  • Improved compilation time for building GPU version
  • Better handling of stray commas in list arguments
  • Added a benchmark that employs Rossman Store Sales dataset to compare quality of GBDT packages
  • Added references to Catboost papers in R-package CITATION file
  • Fixed a build issue in compilation for GPU
  • Fixed a bug in model applicator
  • Fixed model conversion, #533
  • Returned pre 0.11 behaviour for best_score_ and evals_result_ (issue #539)
  • Make valid RECORD in wheel (issue #534)

Release 0.11.0

Changes:

  • Changed default border count for float feature binarization to 254 on CPU to achieve better quality
  • Fixed random seed to 0 by default
  • Support model with more than 254 feature borders or one hot values when doing predictions
  • Added model summation support in python: use catboost.sum_models() to sum models with provided weights.
  • Added json model tutorial json_model_tutorial.ipynb

Release 0.10.4.1

Changes:

  • Bugfix for #518

Release 0.10.4

Breaking changes:

In python 3 some functions returned dictionaries with keys of type bytes - particularly eval_metrics and get_best_score. These are fixed to have keys of type str.

Changes:

  • New metric NumErrors:greater_than=value
  • New metric and objective L_q:q=value
  • model.score(X, y) - can now work with Pool and labels from Pool

Release 0.10.3

Changes:

  • Added EvalResult output after GPU catboost training
  • Supported prediction type option on GPU
  • Added get_evals_result() method and evals_result_ property to model in python wrapper to allow user access metric values
  • Supported string labels for GPU training in cmdline mode
  • Many improvements in JNI wrapper
  • Updated NDCG metric: speeded up and added NDCG with exponentiation in numerator as a new NDCG mode
  • CatBoost doesn't drop unused features from model after training
  • Write training finish time and catboost build info to model metadata
  • Fix automatic pairs generation for GPU PairLogitPairwise target

Release 0.10.2

Main changes:

  • Fixed Python 3 support in catboost.FeaturesData
  • 40% speedup QuerySoftMax CPU training

Release 0.10.1

Improvements

  • 2x Speedup pairwise loss functions
  • For all the people struggling with occasional NaNs in test datasets - now we only write warnings about it

Bugfixes

  • We set up default loss_function in CatBoostClassifier and CatBoostRegressor
  • Catboost write Warning and Error logs to stderr

Release 0.10.0

Breaking changes

R package

  • In R package we have changed parameter name target to label in method save_pool()

Python package

  • We don't support Python 3.4 anymore
  • CatBoostClassifier and CatBoostRegressor get_params() method now returns only the params that were explicitly set when constructing the object. That means that CatBoostClassifier and CatBoostRegressor get_params() will not contain 'loss_function' if it was not specified. This also means that this code:
model1 = CatBoostClassifier()
params = model1.get_params()
model2 = CatBoost(params)

will create model2 with default loss_function RMSE, not with Logloss. This breaking change is done to support sklearn interface, so that sklearn GridSearchCV can work.

  • We've removed several attributes and changed them to functions. This was needed to avoid sklearn warnings: is_fitted_ => is_fitted() metadata_ => get_metadata()
  • We removed file with model from constructor of estimator. This was also done to avoid sklearn warnings.

Educational materials

  • We added tutorial for our ranking modes.
  • We published our slides, you are very welcome to use them.

Improvements

All

  • Now it is possible to save model in json format.
  • We have added Java interface for CatBoost model
  • We now have static linkage with CUDA, so you don't have to install any particular version of CUDA to get catboost working on GPU.
  • We implemented both multiclass modes on GPU, it is very fast.
  • It is possible now to use multiclass with string labels, they will be inferred from data
  • Added use_weights parameter to metrics. By default all metrics, except for AUC use weights, but you can disable it. To calculate metric value without weights, you need to set this parameter to false. Example: Accuracy:use_weights=false. This can be done only for custom_metrics or eval_metric, not for the objective function. Objective function always uses weights if they are present in the dataset.
  • We now use snapshot time intervals. It will work much faster if you save snapshot every 5 or 10 minutes instead of saving it on every iteration.
  • Reduced memory consumption by ranking modes.
  • Added automatic feature importance evaluation after completion of GPU training.
  • Allow inexistent indexes in ignored features list
  • Added new metrics: LogLikelihoodOfPrediction, RecallAt:top=k, PrecisionAt:top=k and MAP:top=k.
  • Improved quality for multiclass with weighted datasets.
  • Pairwise modes now support automatic pairs generation (see tutorial for that).
  • Metric QueryAverage is renamed to a more clear AverageGain. This is a very important ranking metric. It shows average target value in top k documents of a group. Introduced parameter best_model_min_trees - the minimal number of trees the best model should have.

Python

  • We now support sklearn GridSearchCV: you can pass categorical feature indices when constructing estimator. And then use it in GridSearchCV.
  • We added new method to utils - building of ROC curve: get_roc_curve.
  • Added get_gpu_device_count() method to python package. This is a way to check if your CUDA devices are available.
  • We implemented automatical selection of decision-boundary using ROC curve. You can select best classification boundary given the maximum FPR or FNR that you allow to the model. Take a look on catboost.select_threshold(self, data=None, curve=None, FPR=None, FNR=None, thread_count=-1). You can also calculate FPR and FNR for each boundary value.
  • We have added pool slicing: pool.slice(doc_indices)
  • Allow GroupId and SubgroupId specified as strings.

R package

  • GPU support in R package. You need to use parameter task_type='GPU' to enable GPU training.
  • Models in R can be saved/restored by means of R: save/load or saveRDS/readRDS

Speedups

  • New way of loading data in Python using FeaturesData structure. Using FeaturesData will speed up both loading data for training and for prediction. It is especially important for prediction, because it gives around 10 to 20 times python prediction speedup.
  • Training multiclass on CPU ~ 60% speedup
  • Training of ranking modes on CPU ~ 50% speedup
  • Training of ranking modes on GPU ~ 50% speedup for datasets with many features and not very many objects
  • Speedups of metric calculation on GPU. Example of speedup on our internal dataset: training with - AUC eval metric with test dataset with 2kk objects is speeded up 7sec => 0.2 seconds per iteration.
  • Speedup of all modes on CPU training.

We also did a lot of stability improvements, and improved usability of the library, added new parameter synonyms and improved input data validations.

Thanks a lot to all people who created issues on github. And thanks a lot to our contributor @pukhlyakova who implemented many new useful metrics!

Release 0.9.1.1

Bugfixes

  • Fixed #403 bug in cuda train submodule (training crashed without evaluation set)
  • Fixed exception propagation on pool parsing stage
  • Add support of string GroupId and SubgroupId in python-package
  • Print real class names instead of their labels in eval output

Release 0.9

Breaking Changes

  • We removed calc_feature_importance parameter from Python and R. Now feature importance calculation is almost free, so we always calculate feature importances. Previously you could disable it if it was slowing down your training.
  • We removed Doc type for feature importances. Use Shap instead.
  • We moved thread_count parameter in Python get_feature_importance method to the end.

Ranking

In this release we added several very powerfull ranking objectives:

  • PairLogitPairwise
  • YetiRankPairwise
  • QueryCrossEntropy (GPU only)

Other ranking improvements:

  • We have made improvements to our existing ranking objectives QuerySoftMax and PairLogit.
  • We have added group weights support.

Accuracy improvements

  • Improvement for datasets with weights
  • Now we automatically calculate a good learning rate for you in the start of training, you don't have to specify it. After the training has finished, you can look on the training curve on evaluation dataset and make ajustments to the selected learning rate, but it will already be a good value.

Speedups:

  • Several speedups for GPU training.
  • 1.5x speedup for applying the model.
  • Speed up multi classificaton training.
  • 2x speedup for AUC calculation in eval_metrics.
  • Several speedups for eval_metrics for other metrics.
  • 100x speed up for Shap values calculation.
  • Speedup for feature importance calculation. It used to be a bottleneck for GPU training previously, now it's not.
  • We added possibility to not calculate metric on train dataset using MetricName:hints=skip_train~false (it might speed up your training if metric calculation is a bottle neck, for example, if you calculate many metrics or if you calculate metrics on GPU).
  • We added possibility to calculate metrics only periodically, not on all iterations. Use metric_period for that. (previously it only disabled verbose output on each iteration).
  • Now we disable by default calculation of expensive metrics on train dataset. We don't calculate AUC and PFound metrics on train dataset by default. You can also disable calculation of other metrics on train dataset using MetricName:hints=skip_train~true. If you want to calculate AUC or PFound on train dataset you can use MetricName:hints=skip_train~false.
  • Now if you want to calculate metrics using eval_metrics or during training you can use metric_period to skip some iterations. It will speed up eval_metrics and it might speed up training, especially GPU training. Note that the most expensive metric calculation is AUC calculation, for this metric and large datasets it makes sense to use metric_period. If you only want to see less verbose output, and still want to see metric values on every iteration written in file, you can use verbose=n parameter
  • Parallelization of calculation of most of the metrics during training

Improved GPU experience

  • It is possible now to calculate and visualise custom_metric during training on GPU. Now you can use our Jupyter visualization, CatBoost viewer or TensorBoard the same way you used it for CPU training. It might be a bottleneck, so if it slows down your training use metric_period=something and MetricName:hints=skip_train~false
  • We switched to CUDA 9.1. Starting from this release CUDA 8.0 will not be supported
  • Support for external borders on GPU for cmdline

Improved tools for model analysis

  • We added support of feature combinations to our Shap values implementation.
  • Added Shap values for MultiClass and added an example of it's usage to our Shap tutorial.
  • Added pretified parameter to get_feature_importance(). With pretified=True the function will return list of features with names sorted in descending order by their importance.
  • Improved interfaces for eval-feature functionality
  • Shap values support in R-package

New features

  • It is possible now to save any metainformation to the model.
  • Empty values support
  • Better support of sklearn
  • feature_names_ for CatBoost class
  • Added silent parameter
  • Better stdout
  • Better diagnostic for invalid inputs
  • Better documentation
  • Added a flag to allow constant labels

New metrics

We added many new metrics that can be used for visualization, overfitting detection, selecting of best iteration of training or for cross-validation:

  • BierScore
  • HingeLoss
  • HammingLoss
  • ZeroOneLoss
  • MSLE
  • MAE
  • BalancedAccuracy
  • BalancedErrorRate
  • Kappa
  • Wkappa
  • QueryCrossEntropy
  • NDCG

New ways to apply the model

  • Saving model as C++ code
  • Saving model with categorical features as Python code

New ways to build the code

Added make files for binary with CUDA and for Python package

Tutorials

We created a new repo with tutorials, now you don't have to clone the whole catboost repo to run Jupyter notebook with a tutorial.

Bugfixes

We have also a set of bugfixes and we are gratefull to everyone who has filled a bugreport, helping us making the library better.

Thanks to our Contributors

This release contains contributions from CatBoost team. We want to especially mention @pukhlyakova who implemented lots of useful metrics.

Release 0.8.1

Bug Fixes and Other Changes

  • New model method get_cat_feature_indices() in Python wrapper.
  • Minor fixes and stability improvements.

Release 0.8

Breaking changes

  • We fixed bug in CatBoost. Pool initialization from numpy.array and pandas.dataframe with string values that can cause slight inconsistence while using trained model from older versions. Around 1% of cat feature hashes were treated incorrectly. If you expirience quality drop after update you should consider retraining your model.

Major Features And Improvements

  • Algorithm for finding most influential training samples for a given object from the 'Finding Influential Training Samples for Gradient Boosted Decision Trees' paper is implemented. This mode for every object from input pool calculates scores for every object from train pool. A positive score means that the given train object has made a negative contribution to the given test object prediction. And vice versa for negative scores. The higher score modulo - the higher contribution. See get_object_importance model method in Python package and ostr mode in cli-version. Tutorial for Python is available here. More details and examples will be published in documentation soon.
  • We have implemented new way of exploring feature importance - Shap values from paper. This allows to understand which features are most influent for a given object. You can also get more insite about your model, see details in a tutorial.
  • Save model as code functionality published. For now you could save model as Python code with categorical features and as C++ code w/o categorical features.

Bug Fixes and Other Changes

  • Fix _catboost reinitialization issues #268 and #269.
  • Python module catboost.util extended with create_cd. It creates column description file.
  • Now it's possible to load titanic and amazon (Kaggle Amazon Employee Access Challenge) datasets from Python code. Use catboost.datasets.
  • GPU parameter use_cpu_ram_for_cat_features renamed to gpu_cat_features_storage with posible values CpuPinnedMemory and GpuRam. Default is GpuRam.

Thanks to our Contributors

This release contains contributions from CatBoost team.

As usual we are grateful to all who filed issues or helped resolve them, asked and answered questions.

Release 0.7.2

Major Features And Improvements

  • GPU: New DocParallel mode for tasks without categorical features and or with categorical features and —max-ctr-complextiy 1. Provides best performance for pools with big number of documents.
  • GPU: Distributed training on several GPU host via MPI. See instruction how to build binary here.
  • GPU: Up to 30% learning speed-up for Maxwell and later GPUs with binarization level > 32

Bug Fixes and Other Changes

  • Hotfixes for GPU version of python wrapper.

Release 0.7.1

Major Features And Improvements

  • Python wrapper: added methods to download datasets titanic and amazon, to make it easier to try the library (catboost.datasets).
  • Python wrapper: added method to write column desctiption file (catboost.utils.create_cd).
  • Made improvements to visualization.
  • Support non-numeric values in GroupId column.
  • Tutorials section updated.

Bug Fixes and Other Changes

  • Fixed problems with eval_metrics (issue #285)
  • Other fixes

Release 0.7

Breaking changes

  • Changed parameter order in train() function to be consistant with other GBDT libraries.
  • use_best_model is set to True by default if eval_set labels are present.

Major Features And Improvements

  • New ranking mode YetiRank optimizes NDGC and PFound.
  • New visualisation for eval_metrics and cv in Jupyter notebook.
  • Improved per document feature importance.
  • Supported verbose=int: if verbose > 1, metric_period is set to this value.
  • Supported type(eval_set) = list in python. Currently supporting only single eval_set.
  • Binary classification leaf estimation defaults are changed for weighted datasets so that training converges for any weights.
  • Add model_size_reg parameter to control model size. Fix ctr_leaf_count_limit parameter, also to control model size.
  • Beta version of distributed CPU training with only float features support.
  • Add subgroupId to Python/R-packages.
  • Add groupwise metrics support in eval_metrics.

Thanks to our Contributors

This release contains contributions from CatBoost team.

We are grateful to all who filed issues or helped resolve them, asked and answered questions.

Release 0.6.3

Breaking changes

  • boosting_type parameter value Dynamic is renamed to Ordered.
  • Data visualisation functionality in Jupyter Notebook requires ipywidgets 7.x+ now.
  • query_id parameter renamed to group_id in Python and R wrappers.
  • cv returns pandas.DataFrame by default if Pandas installed. See new parameter as_pandas.

Major Features And Improvements

  • CatBoost build with make file. Now it’s possible to build command-line CPU version of CatBoost under Linux with make file.
  • In column description column name Target is changed to Label. It will still work with previous name, but it is recommended to use the new one.
  • eval-metrics mode added into cmdline version. Metrics can be calculated for a given dataset using a previously trained model.
  • New classification metric CtrFactor is added.
  • Load CatBoost model from memory. You can load your CatBoost model from file or initialize it from buffer in memory.
  • Now you can run fit function using file with dataset: fit(train_path, eval_set=eval_path, column_description=cd_file). This will reduce memory consumption by up to two times.
  • 12% speedup for training.

Bug Fixes and Other Changes

  • JSON output data format is changed.
  • Python whl binaries with CUDA 9.1 support for Linux OS published into the release assets.
  • Added bootstrap_type parameter to CatBoostClassifier and Regressor (issue #263).

Thanks to our Contributors

This release contains contributions from newbfg and CatBoost team.

We are grateful to all who filed issues or helped resolve them, asked and answered questions.

Release 0.6.2

Major Features And Improvements

  • BETA version of distributed mulit-host GPU via MPI training
  • Added possibility to import coreml model with oblivious trees. Makes possible to migrate pre-flatbuffers model (with float features only) to current format (issue #235)
  • Added QuerySoftMax loss function

Bug Fixes and Other Changes

  • Fixed GPU models bug on pools with both categorical and float features (issue #241)
  • Use all available cores by default
  • Fixed not querywise loss for pool with QueryId
  • Default float features binarization method set to GreedyLogSum

Release 0.6.1.1

Bug Fixes and Other Changes

  • Hotfix for critical bug in Python and R wrappers (issue #238)
  • Added stratified data split in CV
  • Fix is_classification check and CV for Logloss

Release 0.6.1

Bug Fixes and Other Changes

  • Fixed critical bugs in formula evaluation code (issue #236)
  • Added scale_pos_weight parameter

Release 0.6

Speedups

  • 25% speedup of the model applier
  • 43% speedup for training on large datasets.
  • 15% speedup for QueryRMSE and calculation of querywise metrics.
  • Large speedups when using binary categorical features.
  • Significant (x200 on 5k trees and 50k lines dataset) speedup for plot and stage predict calculations in cmdline.
  • Compilation time speedup.

Major Features And Improvements

  • Industry fastest applier implementation.
  • Introducing new parameter boosting-type to switch between standard boosting scheme and dynamic boosting, described in paper "Dynamic boosting".
  • Adding new bootstrap types bootstrap_type, subsample. Using Bernoulli bootstrap type with subsample < 1 might increase the training speed.
  • Better logging for cross-validation, added parameter logging_level and metric_period (should be set in training parameters) to cv.
  • Added a separate train function that receives the parameters and returns a trained model.
  • Ranking mode QueryRMSE now supports default settings for dynamic boosting.
  • R-package pre-build binaries are included into release.
  • We added many synonyms to our parameter names, now it is more convenient to try CatBoost if you are used to some other library.

Bug Fixes and Other Changes

  • Fix for CPU QueryRMSE with weights.
  • Adding several missing parameters into wrappers.
  • Fix for data split in querywise modes.
  • Better logging.
  • From this release we'll provide pre-build R-binaries
  • More parallelisation.
  • Memory usage improvements.
  • And some other bug fixes.

Thanks to our Contributors

This release contains contributions from CatBoost team.

We are grateful to all who filed issues or helped resolve them, asked and answered questions.

Release 0.5.2

Major Features And Improvements

  • We've made single document formula applier 4 times faster!
  • model.shrink function added in Python and R wrappers.
  • Added new training parameter metric_period that controls output frequency.
  • Added new ranking metric QueryAverage.
  • This version contains an easy way to implement new user metrics in C++. How-to example is provided.

Bug Fixes and Other Changes

  • Stability improvements and bug fixes

As usual we are grateful to all who filed issues, asked and answered questions.

Release 0.5

Breaking Changes

Cmdline:

  • Training parameter gradient-iterations renamed to leaf-estimation-iterations.
  • border option removed. If you want to specify border for binary classification mode you need to specify it in the following way: loss-function Logloss:Border=0.5
  • CTR parameters are changed:
    • Removed priors, per-feature-priors, ctr-binarization;
    • Added simple-ctr, combintations-ctr, per-feature-ctr; More details will be published in our documentation.

Python:

  • Training parameter gradient_iterations renamed to leaf_estimation_iterations.
  • border option removed. If you want to specify border for binary classification mode you need to specify it in the following way: loss_function='Logloss:Border=0.5'
  • CTR parameters are changed:
    • Removed priors, per_feature_priors, ctr_binarization;
    • Added simple_ctr, combintations_ctr, per_feature_ctr; More details will be published in our documentation.

Major Features And Improvements

  • In Python we added a new method eval_metrics: now it's possible for a given model to calculate specified metric values for each iteration on specified dataset.
  • One command-line binary for CPU and GPU: in CatBoost you can switch between CPU and GPU training by changing single parameter value task-type CPU or GPU (task_type 'CPU', 'GPU' in python bindings). Windows build still contains two binaries.
  • We have speed up the training up to 30% for datasets with a lot of objects.
  • Up to 10% speed-up of GPU implementation on Pascal cards

Bug Fixes and Other Changes

  • Stability improvements and bug fixes

As usual we are grateful to all who filed issues, asked and answered questions.

Release 0.4

Breaking Changes

FlatBuffers model format: new CatBoost versions wouldn’t break model compatibility anymore.

Major Features And Improvements

  • Training speedups: we have speed up the training by 33%.
  • Two new ranking modes are available:
    • PairLogit - pairwise comparison of objects from the input dataset. Algorithm maximises probability correctly reorder all dataset pairs.
    • QueryRMSE - mix of regression and ranking. It’s trying to make best ranking for each dataset query by input labels.

Bug Fixes and Other Changes

  • We have fixed a bug that caused quality degradation when using weights < 1.
  • Verbose flag is now deprecated, please use logging_level instead. You could set the following levels: Silent, Verbose, Info, Debug.
  • And some other bugs.

Thanks to our Contributors

This release contains contributions from: avidale, newbfg, KochetovNicolai and CatBoost team.

We are grateful to all who filed issues or helped resolve them, asked and answered questions.

Release 0.3

Major Features And Improvements

GPU CUDA support is available. CatBoost supports multi-GPU training. Our GPU implementation is 2 times faster then LightGBM and more then 20 times faster then XGBoost one. Check out the news with benchmarks on our site.

Bug Fixes and Other Changes

Stability improvements and bug fixes

Thanks to our Contributors

This release contains contributions from: daskol and CatBoost team.

We are grateful to all who filed issues or helped resolve them, asked and answered questions.

Release 0.2

Breaking Changes

  • R library interface significantly changed
  • New model format: CatBoost v0.2 model binary not compatible with previous versions
  • Cross-validation parameters changes: we changed overfitting detector parameters of CV in python so that it is same as those in training.
  • CTR types: MeanValue => BinarizedTargetMeanValue

Major Features And Improvements

  • Training speedups: we have speed up the training by 20-30%.
  • Accuracy improvement with categoricals: we have changed computation of statistics for categorical features, which leads to better quality.
  • New type of overfitting detector: Iter. This type of detector was requested by our users. So now you can also stop training by a simple criterion: if after a fixed number of iterations there is no improvement of your evaluation function.
  • TensorBoard support: this is another way of looking on the graphs of different error functions both during training and after training has finished. To look at the metrics you need to provide train_dir when training your model and then run "tensorboard --logdir={train_dir}"
  • Jupyter notebook improvements: for our Python library users that experiment with Jupyter notebooks, we have improved our visualisation tool. Now it is possible to save image of the graph. We also have changed scrolling behaviour so that it is more convenient to scroll the notebook.
  • NaN features support: we also have added simple but effective way of dealing with NaN features. If you have some NaNs in the train set, they will be changed to a value that is less than the minimum value or greater than the maximum value in the dataset (this is configurable), so that it is guaranteed that they are in their own bin, and a split would separates NaN values from all other values. By default, no NaNs are allowed, so you need to use option nan_mode for that. When applying a model, NaNs will be treated in the same way for the features where NaN values were seen in train. It is not allowed to have NaN values in test if no NaNs in train for this feature were provided.
  • Snapshotting: we have added snapshotting to our Python and R libraries. So if you think that something can happen with your training, for example machine can reboot, you can use snapshot_file parameter - this way after you restart your training it will start from the last completed iteration.
  • R library tutorial: we have added tutorial
  • Logging customization: we have added allow_writing_files parameter. By default some files with logging and diagnostics are written on disc, but you can turn it off using by setting this flag to False.
  • Multiclass mode improvements: we have added a new objective for multiclass mode - MultiClassOneVsAll. We also added class_names param - now you don't have to renumber your classes to be able to use multiclass. And we have added two new metrics for multiclass: TotalF1 and MCC metrics. You can use the metrics to look how its values are changing during training or to use overfitting detection or cutting the model by best value of a given metric.
  • Any delimeters support: in addition to datasets in tsv format, CatBoost now supports files with any delimeters

Bug Fixes and Other Changes

Stability improvements and bug fixes

Thanks to our Contributors

This release contains contributions from: grayskripko, hadjipantelis and CatBoost team.

We are grateful to all who filed issues or helped resolve them, asked and answered questions.