Skip to content
For Managed AI Platform users, see the cloud platform documentation
DataRobot API resources > API reference documentation > API changelogs > Python client changelog > v3.7 changelog

v3.7 changelog

Reference the changes introduced to v3.6 of DataRobot's Python API client.

Python client v3.7

New features

  • The DataRobot Python Client now supports Python 3.12 and Python 3.13.
  • Added Deployment.get_retraining_settings to retrieve retraining settings.
  • Added Deployment.update_retraining_settings to update retraining settings.
  • Updated RESTClientObject to retry requests when the server returns a 104 connection reset error.
  • Added support for datasphere as an intake and output type in batch predictions.
  • Added Deployment.get_accuracy_metrics_settings to retrieve accuracy metrics settings.
  • Added Deployment.update_accuracy_metrics_settings to update accuracy metrics settings.
  • Added CustomMetricValuesOverSpace to retrieve custom metric values over space.
  • Added CustomMetric.get_values_over_space to retrieve custom metric values over space.
  • Created ComplianceDocTemplateProjectType, an enum to define the project type supported by the compliance documentation custom template.
  • Added attribute project_type to ComplianceDocTemplate to identify the template supported project type.
  • Added optional parameter project_type in ComplianceDocTemplate.get_default to retrieve the project type's default template.
  • Added optional parameter project_type in ComplianceDocTemplate.create to specify the project type supported by the template to create.
  • Added optional parameter project_type in ComplianceDocTemplate.create_from_json_file to specify the project type supported by the template to create.
  • Added optional parameter project_type in ComplianceDocTemplate.update to allow updating an existing template's project type.
  • Added optional parameter project_type in ComplianceDocTemplate.list to allow to filtering and searching by a template's project type.
  • Added ShapMatrix.get_as_csv to retrieve SHAP matrix results as a CSV file.
  • Added ShapMatrix.get_as_dataframe to retrieve SHAP matrix results as a dataframe.
  • Added a new class LiftChart to interact with lift chart insights.
  • Added a new class RocCurve to interact with ROC curve insights.
  • Added a new class Residuals to interact with residuals insights.
  • Added Project.create_from_recipe to create Feature Discovery projects using recipes.
  • Added an optional parameter recipe_type to datarobot.models.Recipe.from_dataset to create Wrangling recipes.
  • Added an optional parameter recipe_type to datarobot.models.Recipe.from_data_store to create Wrangling recipes.
  • Added Recipe.set_recipe_metadata to update recipe metadata.
  • Added an optional parameter snapshot_policy to datarobot.models.Recipe.from_dataset to specify the snapshot policy to use.
  • Added a new attribute snapshot_policy to datarobot.models.RecipeDatasetInput to specify the snapshot policy to use.
  • Added a new attribute dataset_id to datarobot.models.JDBCTableDataSourceInput to specify the exact dataset ID to use.
  • Added new attributes prediction_point, relationships_configuration_id and feature_discovery_supervised_feature_reduction to RecipeSettings.
  • Added several optional parameters to ExecutionEnvironment for list, create and update methods.
  • Added optional parameter metadata_filter to ComparisonPrompt.create.
  • Added CustomInferenceModel.share to update access control settings for a custom model.
  • Added CustomInferenceModel.get_access_list to retrieve access control settings for a custom model.
  • Added new attribute latest_successful_version to ExecutionEnvironment.
  • Added Dataset.create_from_project to create datasets from project data.
  • Added ExecutionEnvironment.share to update access control settings for an execution environment.
  • Added ExecutionEnvironment.get_access_list to retrieve access control settings an execution environment.
  • Created ModerationTemplate to interact with LLM moderation templates.
  • Created ModerationConfiguration to interact with LLM moderation configuration.
  • Created CustomTemplate to interact with custom-templates elements.
  • Extended the advanced options available when setting a target to include parameter: feature_engineering_prediction_point (part of the AdvancedOptions object).
  • Added optional parameter substitute_url_parameters to DataStore for list and get methods.
  • Added Model.start_incremental_learning_from_sample to initialize the incremental learning model and begin training using the chunking service. Requires the "Project Creation from a Dataset Sample" feature flag.
  • Added NonChatAwareCustomModelValidation as the base class for CustomModelVectorDatabaseValidation and CustomModelEmbeddingValidation. In contrast, CustomModelLLMValidation now implements the create and update methods differently to interact with the deployments that support the chat completion API.
  • Added optional parameter chat_model_id to CustomModelLLMValidation.create and CustomModelLLMValidation.update to allow adding deployed LLMs that support the chat completion API.
  • Fixed ComparisonPrompt not being able to load errored comparison prompt results.
  • Added optional parameters retirement_date, is_deprecated, and is_active to LLMDefinition and added an optional parameter llm_is_deprecated to the MetricMetadata to expose LLM deprecation and retirement-related information.

Enhancements

  • Added Deployment.share as an alias for Deployment.update_shared_roles.
  • Added the use_tcp_keepalive parameter to Client to enable TCP keep-alive packets when connections are timing out, enabled by default.
  • Internally use the existing input argument max_wait in CustomModelVersion.clean_create, to set the READ request timeout.

Bugfixes

  • Made user_id and username fields in management_meta optional for PredictionEnvironment to support API responses without these fields.
  • Fixed the enum values of ComplianceDocTemplateType.
  • Fixed the enum values of WranglingOperations.
  • Fixed the enum values of DataWranglingDialect.
  • The Playground ID parameter is no longer optional in EvaluationDatasetConfiguration.list.
  • Copy insights path fixed in MetricInsights.copy_to_playground.
  • Missing fields for prompt_type and warning were added to PromptTrace <datarobot.models.genai.prompt_trace.PromptTrace>.
  • Fixed a query parameter name in SidecarModelMetricValidation.list.
  • Fixed a typo in attribute VectorDatabase: metadata_columns which was metada_columns.
  • Do not camelCase metadata_filter dict in ChatPrompt.create.
  • Fixed a Use Case query parameter name in CustomModelLLMValidation.list, CustomModelEmbeddingValidation.list, and CustomModelVectorDatabaseValidation.list.
  • Fixed featureDiscoverySettings parameter name in RelationshipsConfiguration.create and RelationshipsConfiguration.replace.

API changes

  • Method CustomModelLLMValidation.create no longer requires the prompt_column_name and target_column_name parameters, and can accept an optional chat_model_id parameter. The parameter order has changed. If the custom model LLM deployment supports the chat completion API, it is recommended to use chat_model_id now instead of (or in addition to) specifying the column names.

Deprecation summary

  • Removed the deprecated capabilities attribute of Deployment.
  • Method Model.request_lift_chart is deprecated and will be removed in favor of LiftChart.compute.
  • Method Model.get_lift_chart is deprecated and will be removed in favor of LiftChart.get.
  • Method Model.get_all_lift_charts is deprecated and will be removed in favor of LiftChart.list.
  • Method Model.request_roc_curve is deprecated and will be removed in favor of RocCurve.compute.
  • Method Model.get_roc_curve is deprecated and will be removed in favor of RocCurve.get.
  • Method Model.get_all_roc_curves is deprecated and will be removed in favor of RocCurve.list.
  • Method Model.request_residuals_chart is deprecated and will be removed in favor of Residuals.compute.
  • Method Model.get_residuals_chart is deprecated and will be removed in favor of Residuals.get.
  • Method Model.get_all_residuals_charts is deprecated and will be removed in favor of Residuals.list.

Documentation changes

  • Removed numpydoc as a dependency. Docstring parsing has been handled by sphinx.ext.napoleon since 3.6.0.
  • Fix issues with how the Table of Contents is rendered on ReadTheDocs. sphinx-external-toc is now a dev dependency.
  • Fix minor issues with formatting across the ReadTheDocs site.
  • Updated docs on Anomaly Assessment objects to remove duplicate information.

Experimental changes

  • Added use_case and deployment_id properties to RetrainingPolicy class.
  • Added create and update_use_case methods to RetrainingPolicy class.
  • Renamed the method train_first_incremental_from_sample to start_incremental_learning_from_sample. Added new parameters : early_stopping_rounds and first_iteration_only.
  • Added the credentials_id parameter to the create method in ChunkDefinition.
  • Bugfix the next_run_time property of the NotebookScheduledJob class to be nullable.
  • Added the highlight_whitespace property to the NotebookSettings.
  • Create new directory specifically for notebooks in the experimental portion of the client.
  • Added methods to the Notebook class to work with session: start_session(), stop_session(), get_session_status(), is_running().
  • Added methods to the Notebook in order to execute and check related execution status: execute(), get_execution_status(), is_finished_executing().
  • Added Notebook.create_revision to the Notebook class in order to create revisions.
  • Moved ModerationTemplate class to ModerationTemplate.
  • Moved ModerationConfiguration class to ModerationConfiguration to interact with LLM moderation configuration.
  • Updates to Notebook.run method in the Notebook class in order to encourage proper usage as well as add more descriptive TypedDict as annotation.
  • Added NotebookScheduledJob.get_most_recent_run to the NotebookScheduledJob class to aid in more idiomatic code when dealing with manual runs.
  • Updates to Notebook.run method in the Notebook class in order to support Codespace Notebook execution as well as multiple related new classes and methods to expand API coverage which is needed for the underlying execution.
  • Added ExecutionEnvironment.assign_environment to the ExecutionEnvironment class, which gives the ability to assign or update a notebook's environment.
  • Removed deprecated experimental method Model.get_incremental_learning_metadata.
  • Removed deprecated experimental method Model.start_incremental_learning.