airflow.providers.google.cloud.operators.vertex_ai.auto_ml¶

This module contains Google Vertex AI operators.

Classes¶

`AutoMLTrainingJobBaseOperator`	The base class for operators that launch AutoML jobs on VertexAI.
`CreateAutoMLForecastingTrainingJobOperator`	Create AutoML Forecasting Training job.
`CreateAutoMLImageTrainingJobOperator`	Create Auto ML Image Training job.
`CreateAutoMLTabularTrainingJobOperator`	Create Auto ML Tabular Training job.
`CreateAutoMLVideoTrainingJobOperator`	Create Auto ML Video Training job.
`DeleteAutoMLTrainingJobOperator`	Delete an AutoML training job.
`ListAutoMLTrainingJobOperator`	List an AutoML training job.

Module Contents¶

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.AutoMLTrainingJobBaseOperator(*, project_id, region, display_name, labels=None, parent_model=None, is_default_version=None, model_version_aliases=None, model_version_description=None, training_encryption_spec_key_name=None, model_encryption_spec_key_name=None, training_fraction_split=None, test_fraction_split=None, model_display_name=None, model_labels=None, sync=True, gcp_conn_id='google_cloud_default', impersonation_chain=None, **kwargs)[source]¶

Bases: airflow.providers.google.cloud.operators.cloud_base.GoogleCloudBaseOperator

The base class for operators that launch AutoML jobs on VertexAI.

project_id[source]¶

region[source]¶

display_name[source]¶

labels = None[source]¶

parent_model = None[source]¶

is_default_version = None[source]¶

model_version_aliases = None[source]¶

model_version_description = None[source]¶

training_encryption_spec_key_name = None[source]¶

model_encryption_spec_key_name = None[source]¶

training_fraction_split = None[source]¶

test_fraction_split = None[source]¶

model_display_name = None[source]¶

model_labels = None[source]¶

sync = True[source]¶

gcp_conn_id = 'google_cloud_default'[source]¶

impersonation_chain = None[source]¶

hook: airflow.providers.google.cloud.hooks.vertex_ai.auto_ml.AutoMLHook | None = None[source]¶

property extra_links_params: dict[str, Any][source]¶

Override this method to include parameters for link formatting in extra links.

For example; most of the links on the Google provider require project_id and location in the Link. To be not repeat; you can override this function and return something like the following:

{
    "project_id": self.project_id,
    "location": self.location,
}

on_kill()[source]¶

Act as a callback called when the operator is killed; cancel any running job.

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.CreateAutoMLForecastingTrainingJobOperator(*, dataset_id, target_column, time_column, time_series_identifier_column, unavailable_at_forecast_columns, available_at_forecast_columns, forecast_horizon, data_granularity_unit, data_granularity_count, display_name, model_display_name=None, optimization_objective=None, column_specs=None, column_transformations=None, validation_fraction_split=None, predefined_split_column_name=None, weight_column=None, time_series_attribute_columns=None, context_window=None, export_evaluated_data_items=False, export_evaluated_data_items_bigquery_destination_uri=None, export_evaluated_data_items_override_destination=False, quantiles=None, validation_options=None, budget_milli_node_hours=1000, region, impersonation_chain=None, parent_model=None, window_stride_length=None, window_max_count=None, holiday_regions=None, **kwargs)[source]¶

Bases: AutoMLTrainingJobBaseOperator

Create AutoML Forecasting Training job.

template_fields = ('parent_model', 'dataset_id', 'region', 'impersonation_chain', 'display_name', 'model_display_name')[source]¶

operator_extra_links[source]¶

dataset_id[source]¶

target_column[source]¶

time_column[source]¶

time_series_identifier_column[source]¶

unavailable_at_forecast_columns[source]¶

available_at_forecast_columns[source]¶

forecast_horizon[source]¶

data_granularity_unit[source]¶

data_granularity_count[source]¶

optimization_objective = None[source]¶

column_specs = None[source]¶

column_transformations = None[source]¶

validation_fraction_split = None[source]¶

predefined_split_column_name = None[source]¶

weight_column = None[source]¶

time_series_attribute_columns = None[source]¶

context_window = None[source]¶

export_evaluated_data_items = False[source]¶

export_evaluated_data_items_bigquery_destination_uri = None[source]¶

export_evaluated_data_items_override_destination = False[source]¶

quantiles = None[source]¶

validation_options = None[source]¶

budget_milli_node_hours = 1000[source]¶

window_stride_length = None[source]¶

window_max_count = None[source]¶

holiday_regions = None[source]¶

execute(context)[source]¶

Derive when creating an operator.

The main method to execute the task. Context is the same dictionary used as when rendering jinja templates.

Refer to get_template_context for more context.

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.CreateAutoMLImageTrainingJobOperator(*, dataset_id, prediction_type='classification', multi_label=False, model_type='CLOUD', base_model=None, validation_fraction_split=None, training_filter_split=None, validation_filter_split=None, test_filter_split=None, budget_milli_node_hours=None, disable_early_stopping=False, region, impersonation_chain=None, parent_model=None, **kwargs)[source]¶

Bases: AutoMLTrainingJobBaseOperator

Create Auto ML Image Training job.

template_fields = ('parent_model', 'dataset_id', 'region', 'impersonation_chain')[source]¶

operator_extra_links[source]¶

dataset_id[source]¶

prediction_type = 'classification'[source]¶

multi_label = False[source]¶

model_type = 'CLOUD'[source]¶

base_model = None[source]¶

validation_fraction_split = None[source]¶

training_filter_split = None[source]¶

validation_filter_split = None[source]¶

test_filter_split = None[source]¶

budget_milli_node_hours = None[source]¶

disable_early_stopping = False[source]¶

execute(context)[source]¶

Derive when creating an operator.

The main method to execute the task. Context is the same dictionary used as when rendering jinja templates.

Refer to get_template_context for more context.

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.CreateAutoMLTabularTrainingJobOperator(*, dataset_id, target_column, optimization_prediction_type, optimization_objective=None, column_specs=None, column_transformations=None, optimization_objective_recall_value=None, optimization_objective_precision_value=None, validation_fraction_split=None, predefined_split_column_name=None, timestamp_split_column_name=None, weight_column=None, budget_milli_node_hours=1000, disable_early_stopping=False, export_evaluated_data_items=False, export_evaluated_data_items_bigquery_destination_uri=None, export_evaluated_data_items_override_destination=False, region, impersonation_chain=None, parent_model=None, **kwargs)[source]¶

Bases: AutoMLTrainingJobBaseOperator

Create Auto ML Tabular Training job.

template_fields = ('parent_model', 'dataset_id', 'region', 'impersonation_chain')[source]¶

operator_extra_links[source]¶

dataset_id[source]¶

target_column[source]¶

optimization_prediction_type[source]¶

optimization_objective = None[source]¶

column_specs = None[source]¶

column_transformations = None[source]¶

optimization_objective_recall_value = None[source]¶

optimization_objective_precision_value = None[source]¶

validation_fraction_split = None[source]¶

predefined_split_column_name = None[source]¶

timestamp_split_column_name = None[source]¶

weight_column = None[source]¶

budget_milli_node_hours = 1000[source]¶

disable_early_stopping = False[source]¶

export_evaluated_data_items = False[source]¶

export_evaluated_data_items_bigquery_destination_uri = None[source]¶

export_evaluated_data_items_override_destination = False[source]¶

execute(context)[source]¶

Derive when creating an operator.

The main method to execute the task. Context is the same dictionary used as when rendering jinja templates.

Refer to get_template_context for more context.

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.CreateAutoMLVideoTrainingJobOperator(*, dataset_id, prediction_type='classification', model_type='CLOUD', training_filter_split=None, test_filter_split=None, region, impersonation_chain=None, parent_model=None, **kwargs)[source]¶

Bases: AutoMLTrainingJobBaseOperator

Create Auto ML Video Training job.

template_fields = ('parent_model', 'dataset_id', 'region', 'impersonation_chain')[source]¶

operator_extra_links[source]¶

dataset_id[source]¶

prediction_type = 'classification'[source]¶

model_type = 'CLOUD'[source]¶

training_filter_split = None[source]¶

test_filter_split = None[source]¶

execute(context)[source]¶

Derive when creating an operator.

The main method to execute the task. Context is the same dictionary used as when rendering jinja templates.

Refer to get_template_context for more context.

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.DeleteAutoMLTrainingJobOperator(*, training_pipeline_id, region, project_id, retry=DEFAULT, timeout=None, metadata=(), gcp_conn_id='google_cloud_default', impersonation_chain=None, **kwargs)[source]¶

Bases: airflow.providers.google.cloud.operators.cloud_base.GoogleCloudBaseOperator

Delete an AutoML training job.

Can be used with AutoMLForecastingTrainingJob, AutoMLImageTrainingJob, AutoMLTabularTrainingJob, AutoMLTextTrainingJob, or AutoMLVideoTrainingJob.

template_fields = ('training_pipeline_id', 'region', 'project_id', 'impersonation_chain')[source]¶

training_pipeline_id[source]¶

region[source]¶

project_id[source]¶

retry[source]¶

timeout = None[source]¶

metadata = ()[source]¶

gcp_conn_id = 'google_cloud_default'[source]¶

impersonation_chain = None[source]¶

execute(context)[source]¶

Derive when creating an operator.

The main method to execute the task. Context is the same dictionary used as when rendering jinja templates.

Refer to get_template_context for more context.

class airflow.providers.google.cloud.operators.vertex_ai.auto_ml.ListAutoMLTrainingJobOperator(*, region, project_id, page_size=None, page_token=None, filter=None, read_mask=None, retry=DEFAULT, timeout=None, metadata=(), gcp_conn_id='google_cloud_default', impersonation_chain=None, **kwargs)[source]¶

Bases: airflow.providers.google.cloud.operators.cloud_base.GoogleCloudBaseOperator

List an AutoML training job.

Can be used with AutoMLForecastingTrainingJob, AutoMLImageTrainingJob, AutoMLTabularTrainingJob, AutoMLTextTrainingJob, or AutoMLVideoTrainingJob in a Location.

template_fields = ('region', 'project_id', 'impersonation_chain')[source]¶

operator_extra_links[source]¶

region[source]¶

project_id[source]¶

page_size = None[source]¶

page_token = None[source]¶

filter = None[source]¶

read_mask = None[source]¶

retry[source]¶

timeout = None[source]¶

metadata = ()[source]¶

gcp_conn_id = 'google_cloud_default'[source]¶

impersonation_chain = None[source]¶

property extra_links_params: dict[str, Any][source]¶

Override this method to include parameters for link formatting in extra links.

For example; most of the links on the Google provider require project_id and location in the Link. To be not repeat; you can override this function and return something like the following:

{
    "project_id": self.project_id,
    "location": self.location,
}

execute(context)[source]¶

Derive when creating an operator.

The main method to execute the task. Context is the same dictionary used as when rendering jinja templates.

Refer to get_template_context for more context.