mlflow.xgboost

The mlflow.xgboost module provides an API for logging and loading XGBoost models. This module exports XGBoost models with the following flavors:

XGBoost (native) format

This is the main flavor that can be loaded back into XGBoost.

mlflow.pyfunc

Produced for use by generic pyfunc-based deployment tools and batch inference.

mlflow.xgboost.autolog(importance_types=['weight'])

Note

Experimental: This method may change or be removed in a future release without warning.

Enables automatic logging from XGBoost to MLflow. Logs the following.

• parameters specified in xgboost.train.

• metrics on each iteration (if evals specified).

• metrics at the best iteration (if early_stopping_rounds specified).

• feature importance as JSON files and plots.

• trained model.

Note that the scikit-learn API is not supported.

Parameters

importance_types – importance types to log.

mlflow.xgboost.get_default_conda_env()

Returns

The default Conda environment for MLflow Models produced by calls to save_model() and log_model().

mlflow.xgboost.load_model(model_uri)

Load an XGBoost model from a local file or a run.

Parameters

model_uri –

The location, in URI format, of the MLflow model. For example:

• /Users/me/path/to/local/model

• relative/path/to/local/model

• s3://my_bucket/path/to/model

• runs:/<mlflow_run_id>/run-relative/path/to/model

For more information about supported URI schemes, see Referencing Artifacts.

Returns

An XGBoost model (an instance of xgboost.Booster)

mlflow.xgboost.log_model(xgb_model, artifact_path, conda_env=None, registered_model_name=None, **kwargs)

Log an XGBoost model as an MLflow artifact for the current run.

Parameters

• xgb_model – XGBoost model (an instance of xgboost.Booster) to be saved. Note that models that implement the scikit-learn API are not supported.

• artifact_path – Run-relative artifact path.

• conda_env –

  Either a dictionary representation of a Conda environment or the path to a Conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, the default get_default_conda_env() environment is added to the model. The following is an example dictionary representation of a Conda environment:

  {
      'name': 'mlflow-env',
      'channels': ['defaults'],
      'dependencies': [
          'python=3.7.0',
          'pip': [
              'xgboost==0.90'
          ]
      ]
  }
  

• registered_model_name – Note:: Experimental: This argument may change or be removed in a future release without warning. If given, create a model version under registered_model_name, also creating a registered model if one with the given name does not exist.

• kwargs – kwargs to pass to xgboost.Booster.save_model method.

mlflow.xgboost.save_model(xgb_model, path, conda_env=None, mlflow_model=<mlflow.models.Model object>)

Save an XGBoost model to a path on the local file system.

Parameters

• xgb_model – XGBoost model (an instance of xgboost.Booster) to be saved. Note that models that implement the scikit-learn API are not supported.

• path – Local path where the model is to be saved.

• conda_env –

  Either a dictionary representation of a Conda environment or the path to a Conda environment yaml file. If provided, this describes the environment this model should be run in. At minimum, it should specify the dependencies contained in get_default_conda_env(). If None, the default get_default_conda_env() environment is added to the model. The following is an example dictionary representation of a Conda environment:

  {
      'name': 'mlflow-env',
      'channels': ['defaults'],
      'dependencies': [
          'python=3.7.0',
          'pip': [
              'xgboost==0.90'
          ]
      ]
  }
  

• mlflow_model – mlflow.models.Model this flavor is being added to.