---
metadata:
  - name: generator
    content: Diplodoc Platform v5.50.3
alternate:
  - https://catboost.ai/docs/en/features/feature-importances-calculation.md
---
> **Documentation Index:** Fetch the complete configuration index at https://catboost.ai/docs/en/llms.txt

# Feature importances

<!-- source: en/_includes/work_src/reusage-common-phrases/feature-importance__intro.md -->
CatBoost provides different types of feature importance calculation:

| Feature importance calculation type | Implementations|
|-------------------------------------|----------------|
|The most important features in the formula| - [PredictionValuesChange](../concepts/fstr.md#regular-feature-importance) <br> - [LossFunctionChange](../concepts/fstr.md#regular-feature-importances__lossfunctionchange) <br> - [InternalFeatureImportance](../concepts/fstr.md#internal-feature-importance)|
|The contribution of each feature to the formula| [ShapValues](../concepts/shap-values.md)|
|The features that work well together| - [Interaction](../concepts/feature-interaction.md#feature-interaction-strength)<br>- [InternalInteraction](../concepts/feature-interaction.md#internal-feature-interaction-strength)|
<!-- endsource: en/_includes/work_src/reusage-common-phrases/feature-importance__intro.md -->

<!-- source: en/_includes/work_src/reusage-common-phrases/choose-the-required-implementation-for-more-details.md -->
Choose the implementation for more details.
<!-- endsource: en/_includes/work_src/reusage-common-phrases/choose-the-required-implementation-for-more-details.md -->


## Python package

<!-- source: en/_includes/work_src/reusage-common-phrases/use-one-of-the-following-methods-to-calculate-feature-importances.md -->
Use one of the following methods:
<!-- endsource: en/_includes/work_src/reusage-common-phrases/use-one-of-the-following-methods-to-calculate-feature-importances.md -->

- <!-- source: en/_includes/work_src/reusage-common-phrases/fit__attributes__feature-importance_.md -->
  Use the feature_importances_ attribute.
  <!-- endsource: en/_includes/work_src/reusage-common-phrases/fit__attributes__feature-importance_.md -->

- <!-- source: en/_includes/work_src/reusage-common-phrases/methods__calc-fstr.md -->
  Use one of the following methods to calculate the feature importances after model training:
  <!-- endsource: en/_includes/work_src/reusage-common-phrases/methods__calc-fstr.md -->

|Class|Description|
|-------------------------------------------|-------------------------------------------------|
|[CatBoost](../concepts/python-reference_catboost.md)|[get_feature_importance](../concepts/python-reference_catboost_get_feature_importance.md)|
|[CatBoostClassifier](../concepts/python-reference_catboostclassifier.md)| [get_feature_importance](../concepts/python-reference_catboostclassifier_get_feature_importance.md)|
| [CatBoostRegressor](../concepts/python-reference_catboostregressor.md)| [get_feature_importance](../concepts/python-reference_catboostregressor_get_feature_importance.md)|

These methods calculate and return the [feature importances](https://catboost.ai/docs/en/concepts/fstr.md).

## R package

<!-- source: en/_includes/work_src/reusage-common-phrases/use-one-of-the-following-methods-to-calculate-feature-importances.md -->
Use one of the following methods:
<!-- endsource: en/_includes/work_src/reusage-common-phrases/use-one-of-the-following-methods-to-calculate-feature-importances.md -->


- Use the `feature_importances`[attribute](https://catboost.ai/docs/en/concepts/r-reference.md#attributes) to get the feature importances.

- <!-- source: en/_includes/work_src/reusage-common-phrases/methods__calc-fstr.md -->
  Use one of the following methods to calculate the feature importances after model training:
  <!-- endsource: en/_includes/work_src/reusage-common-phrases/methods__calc-fstr.md -->

|Method|Description|
|-------------------------------------------|-------------------------------------------------|
|[catboost.get_feature_importance](../concepts/r-reference_catboost-get_feature_importance.md)|Calculate the [feature importances](../concepts/fstr.md) ([Feature importance](../concepts/output-data_feature-analysis_feature-importance.md) and [Feature interaction strength](../concepts/output-data_feature-analysis_feature-interaction-strength.md)).|



## Command-line version

Use the following command to calculate the feature importances during model training:

|Command| Command keys | Key description |
|-------------------------------------------|-------------------------------------------------| -------------------------------------------------|
| [catboost fit](../references/training-parameters/index.md)| `--fstr-file` |  The name of the resulting file that contains [regular feature importance](../concepts/output-data_feature-analysis_feature-importance.md#per-feature-importance) data (see [Feature importance](../concepts/fstr.md)). <br> Set the required file name for further feature importance analysis.|
|                                                         |`--fstr-internal-file`|  The name of the resulting file that contains [internal feature importance](../concepts/output-data_feature-analysis_feature-importance.md#internal-feature-importance) data (see [Feature importance](../concepts/fstr.md)). <br> Set the required file name for further internal feature importance analysis.|

Use the following command to calculate the feature importances after model training:

|Command| Purpose|
|-------------------------------------------|-------------------------------------------------|
|[catboost fstr](../concepts/cli-reference_fstr-calc.md)| Calculate feature importances.|

## Related information

[Model analysis](https://catboost.ai/docs/en/concepts/model-analysis.md)