Skip to content

Documentation for Model and related classes

Deployment

The object model represents a deployment request in the Engine

delete()

Delete the deployment

predict(data_source, explain=False)

Run online prediction

Parameters:

Name Type Description Default
data_source DataSource

input data to run the prediction

required
explain bool

set to True of the prediction explanation is expected. Defaults to False.

False

Returns:

Type Description
any

Prediction result and explanations if explain = True

Evaluation

The object model represents a model evaluation request and result in the Engine

EvaluationResult dataclass

Result of model evaluation

details: dict class-attribute

Evaluation result details

summary: dict class-attribute

Evaluation summary

Model

The object model represents a project in the Engine

delete()

Delete the model.

Warning

This action cannot be undone. All trained model data will be deleted.

deploy(name=None, description=None, endpoint_name=None, endpoint_id=None, endpoint_description=None, training_id=None, type='cloud', auto_activate=True, auto_deploy=False, timeout=DEFAULT_TIMEOUT)

Deploy the model

Parameters:

Name Type Description Default
name str

name of the deployment. Defaults to None.

None
description str

description of the deployment. Defaults to None.

None
endpoint_name str

name of the endpoint. Defaults to None.

None
endpoint_id str

ID of the endpoint. Defaults to None.

None
endpoint_description str

description of the endpoint. Defaults to None.

None
training_id str

ID of the training. Defaults to None.

None
type str

description. Defaults to 'cloud'.

'cloud'
auto_activate bool

description. Defaults to True.

True
auto_deploy bool

description. Defaults to False.

False
timeout _type_

description. Defaults to DEFAULT_TIMEOUT.

DEFAULT_TIMEOUT

Returns:

Name Type Description
Deployment Deployment

the object represents the deployment

enable_continuous_learning(enabled)

Enable/disable continuous learning for the model

Parameters:

Name Type Description Default
enabled bool

enabled if True

required

evaluate(train_dataset_version=LATEST_VERSION, test_dataset_id=None, test_dataset_version=LATEST_VERSION, timeout=DEFAULT_TIMEOUT)

Evaluate the model with a test dataset

Parameters:

Name Type Description Default
train_dataset_version int

the dataset version on which the model is trained. Defaults to LATEST_VERSION.

LATEST_VERSION
test_dataset_version int

the version of test dataset. Defaults to LATEST_VERSION.

LATEST_VERSION
timeout int

time to wait for the evaluation processed. Defaults to DEFAULT_TIMEOUT.

DEFAULT_TIMEOUT

Raises:

Type Description
RuntimeError

if the model has not been evaluated on the provided version of test dataset

Returns:

Type Description
Evaluation

the Evaluation object

generate_insights(dataset_version=LATEST_VERSION, timeout=DEFAULT_TIMEOUT)

Generate model insights

Parameters:

Name Type Description Default
dataset_version int

the dataset version on which the model trained. Defaults to LATEST_VERSION.

LATEST_VERSION
timeout int

time (in seconds) to wait for the model insights generated. Defaults to DEFAULT_TIMEOUT.

DEFAULT_TIMEOUT

Raises:

Type Description
RuntimeError

raised if the pipeline was failed

Return

List of all generated artifacts

get_insight(type, dataset_version=LATEST_VERSION)

Get insights

Parameters:

Name Type Description Default
type ModelInsightsType

description

required
dataset_version int

description. Defaults to LATEST_VERSION.

LATEST_VERSION

Returns:

Name Type Description
ModelInsights ModelInsight

description

get_training(dataset_version=LATEST_VERSION)

Get the training on dataset version

Parameters:

Name Type Description Default
dataset_version int

version of the training dataset. Defaults to LATEST_VERSION.

LATEST_VERSION

name() property

Name of the model

Returns:

Type Description
str

Name of the model

run_batch_prediction(data_source, train_dataset_version=LATEST_VERSION, timeout=DEFAULT_TIMEOUT)

Run a batch prediction

Parameters:

Name Type Description Default
data_source DataSource

source of the input data

required
train_dataset_version int

Version of the train dataset on which the model trained. Defaults to LATEST_VERSION.

LATEST_VERSION
timeout int

Wait maximum timeout seconds. Defaults to DEFAULT_TIMEOUT.

DEFAULT_TIMEOUT

Raises:

Type Description
RuntimeError

description

Returns:

Name Type Description
Prediction Prediction

Prediction resource

update(name)

Update model information

Parameters:

Name Type Description Default
name str

updated model name

required
description str

updated model description. Defaults to None.

required

wait_for_trained(dataset_version=LATEST_VERSION, timeout=DEFAULT_TIMEOUT * 2)

Wait until the model trained

Parameters:

Name Type Description Default
dataset_version int

the version of dataset on which the model was trained. Defaults to LATEST_VERSION.

LATEST_VERSION
timeout int

time to wait for the model trained. Defaults to DEFAULT_TIMEOUT*2.

DEFAULT_TIMEOUT * 2

Prediction

The object model represents a prediction request in the Engine