Staged prediction

CatBoost allows to apply a trained model and calculate the results for each i-th tree of the model taking into consideration only the trees in the range [0; i).

Choose the implementation for more details.

Python package

Classes

CatBoost

Class purpose

Training and applying models.

Method

staged_predict

Description

Apply the model to the given dataset and calculate the results taking into consideration only the trees in the range [0; i).

CatBoostRegressor

Class purpose

Training and applying models for the regression problems. When using the applying methods only the predicted class is returned. Provides compatibility with the scikit-learn tools.

Method

staged_predict

Description

Apply the model to the given dataset and calculate the results taking into consideration only the trees in the range [0; i).

CatBoostClassifier

Class purpose

Training and applying models for the classification problems. Provides compatibility with the scikit-learn tools.

Methods

staged_predict

Description

Apply the model to the given dataset and calculate the results taking into consideration only the trees in the range [0; i).

staged_predict_proba

Class purpose

The same as staged_predict with the difference that the results are probabilities that the object belongs to the positive class.

R package

For the catboost.staged_predict method:

Purpose

Apply the model to the given dataset and calculate the results for the specified trees only.

Command-line version

For the catboost calc command:

Purpose

Apply the model.

Command keys

--eval-period

Key description

To reduce the number of trees to use when the model is applied or the metrics are calculated, set the step of the trees to use to eval-period.

This parameter defines the step to iterate over the range [--ntree-start; --ntree-end). For example, let's assume that the following parameter values are set:

  • --ntree-start is set 0
  • --ntree-end is set to N (the total tree count)
  • --eval-period is set to 2

In this case, the results are returned for the following tree ranges: [0, 2), [0, 4), ... , [0, N).