Skip to content

AWSCompatibleCredentials

AWSCompatibleCredentials()

AWS-compatible integration credentials.

Attributes

  • credentials_id : str
  • name : str
  • default : bool
  • type : ExternalIntegration
  • access_key_id : The access key id
  • endpoint_url : The endpoint url (if any)

AWSCredentials

AWSCredentials()

AWS integration credentials.

Attributes

  • credentials_id : str
  • name : str
  • default : bool
  • type : ExternalIntegration
  • role_arn : The ARN of the role that should be assumed via STS

AWSEventBridgeRetrainTrigger

AWSEventBridgeRetrainTrigger()

Base model to define an AWS EventBridge retrain trigger

Fields: type credentials_id aws_region_name event_bus_name

ApiKey

ApiKey()

base model for api key

Attributes

  • api_key : str
  • name : str
  • expiration_time : str | None

AzureBlobDataSource

AzureBlobDataSource()

A source that identifies a blob in Azure Storage.

Attributes

  • file_type : FileType
  • is_folder : bool
  • folder_type : FolderType | None
  • credentials_id : The id of the credentials to use to authenticate to the remote data source. If None, the default will be used
  • object_path : str

Methods:

.get_path

.get_path()

.get_source_type

.get_source_type()

AzureCredentials

AzureCredentials()

Azure integration credentials.

Attributes

  • app_id : The id of the service principal

AzureEventGridRetrainTrigger

AzureEventGridRetrainTrigger()

Base model to define an Azure EventGrid retrain trigger

Fields: type credentials_id topic_endpoint

BinaryClassificationTaskCostInfo

BinaryClassificationTaskCostInfo()

Binary classification cost info is expressed in two terms: - cost of false positive - cost of false negative

CategoricalSegmentRule

CategoricalSegmentRule()

Rule for a segment over categorical values. It contains a list of values that are considered in OR logic to define the rule. See SegmentRule for additional details.

Attributes

  • values : list[str | int]

Methods:

.get_supported_data_types

.get_supported_data_types()

ColumnInfo

ColumnInfo()

Column base model

Attributes

  • name : str
  • role : ColumnRole
  • is_nullable : bool
  • data_type : DataType
  • predicted_target : Optional[str] = None
  • possible_values : Optional[list[str | int | bool]] = None
  • model_id : Optional[str] = None
  • dims : Optional[tuple[int]] = None it is mandatory when data_type is Array
  • classes_names : Optional[list[str]] = None it is mandatory when the column is the target in multilabel classification tasks
  • subrole : Optional[ColumnSubRole] = None Indicates the subrole of the column. It's used in RAG tasks to define the role of the input columns (e.g. user input or retrieved context)
  • image_mode : Optional[ImageMode] = None Indicates the mode of the image. It must be provided when the data type is an image

Company

Company()

Company model

Attributes

  • company_id : str
  • name : str
  • address : str
  • vat : str

CompanyUser

CompanyUser()

base model for company user

Attributes

  • user_id : str
  • company_role : UserCompanyRole

Data

Data()

Generic data model that contains all information about a data

Attributes

  • data_structure : DataStructure
  • source : DataSource

DataBatch

DataBatch()

A Data Batch represents a portion of data that is sent to the ML cube Platform.

Attributes

  • index : int The index of the data batch, assigned in the order of creation
  • creation_date : datetime The creation date of the data batch
  • first_sample_date : datetime The date of the first sample in the data batch
  • last_sample_date : datetime The date of the last sample in the data batch
  • storing_data_type : StoringDataType The origin of the data batch
  • inputs : bool Whether the data batch contains inputs
  • metadata : bool Whether the data batch contains metadata
  • target : bool Whether the data batch contains the target
  • predictions : list[str] The list of models for which the data batch contains predictions
  • monitoring_flags : list[DataBatchMonitoringFlag] The list of monitoring flags referring to the whole population
  • segmented_monitoring_flags : list[SegmentedMonitoringFlags] The list of monitoring flags for each segment

DataBatchMonitoringFlag

DataBatchMonitoringFlag()

Model that stores the monitoring status of a monitoring target, used in the context of a data batch.

Attributes

  • monitoring_target : MonitoringTarget
  • status : MonitoringStatus | None The status of the monitoring target. If None, it means that the monitoring target was not monitored.

DataSchema

DataSchema()

Data schema base model

Attributes

  • columns : List[ColumnInfo]

DataSource

DataSource()

Generic data source.

Attributes

  • file_type : FileType
  • is_folder : bool
  • folder_type : FolderType | None

DetectionEvent

DetectionEvent()

An event created during the detection process.

Attributes

  • event_id : str
  • event_type : DetectionEventType
  • monitoring_target : MonitoringTarget
  • monitoring_metric : MonitoringMetric | None
  • severity_type : Optional[DetectionEventSeverity]
  • insert_datetime : str
  • sample_timestamp : float
  • sample_customer_id : str
  • model_id : Optional[str]
  • model_name : Optional[str]
  • model_version : Optional[str]
  • user_feedback : Optional[bool]
  • specification : Optional[str]
  • segment_id : Optional[str]

DetectionEventAction

DetectionEventAction()

Generic action that can be performed

Attributes

  • type : DetectionEventActionType

DetectionEventRule

DetectionEventRule(
   **kwargs
)

A rule that can be triggered by a detection event, and executes a series of actions.

Attributes

  • rule_id : str
  • name : str
  • task_id : str
  • model_name : Optional[str]
  • severity : DetectionEventSeverity
  • detection_event_type : DetectionEventType
  • monitoring_target : MonitoringTarget
  • actions : List[DetectionEventAction]
  • segment_id : Optional[str]

DiscordNotificationAction

DiscordNotificationAction()

Action that sends a notification to a Discord server through a webhook that you configure

Attributes

  • webhook : str type = DetectionEventActionType.DISCORD_NOTIFICATION

EmailNotificationAction

EmailNotificationAction()

Base Model for Email Notification Action

Attributes

  • address : str type = DetectionEventActionType.EMAIL_NOTIFICATION

EmbeddingData

EmbeddingData()

Embedding data model i.e., a data that can be represented via DataFrame and is stored in formats like: csv, parquet, json. There is only one input that has type array_1

Attributes

  • data_structure : DataStructure = DataStructure.EMBEDDING
  • source : DataSource

GCPCredentials

GCPCredentials()

GCP integration credentials.

Attributes

  • credentials_id : str
  • name : str
  • default : bool
  • type : ExternalIntegration
  • gcp_project_id : The id of the project on GCP
  • client_email : The email that identifies the service account
  • client_id : The client id

GCPPubSubRetrainTrigger

GCPPubSubRetrainTrigger()

Base model to define a GCP PubSub retrain trigger

Fields: type credentials_id topic_name

GCSDataSource

GCSDataSource()

A source that identifies a file in a GCS bucket.

Attributes

  • file_type : FileType
  • is_folder : bool
  • folder_type : FolderType | None
  • credentials_id : The id of the credentials to use to authenticate to the remote data source. If None, the default will be used
  • object_path : str

Methods:

.get_path

.get_path()

.get_source_type

.get_source_type()

ImageData

ImageData()

Image data model i.e., images, text or other. Since it is composed of multiple files, it needs a mapping between customer ids and those files

Attributes

  • data_structure : DataStructure = DataStructure.IMAGE
  • source : DataSource
  • mapping_source : DataSource
  • embedding_source : DataSource | None

IntegrationCredentials

IntegrationCredentials()

Credentials to authenticate to a 3rd party service provider via an integration.

Attributes

  • credentials_id : str
  • name : str
  • default : bool
  • type : ExternalIntegration

Job

Job()

Job information item model

Attributes

  • job_id : str
  • job_group : str
  • project_id : str
  • project_name : str
  • task_id : str
  • task_name : str
  • model_id : Optional[str]
  • model_name : Optional[str]
  • status : str
  • error : Optional[str]

KPI

KPI()

KPI base model

Attributes

  • kpi_id : str
  • name : str
  • status : ModelStatus
  • status_kpi_start_timestamp : Optional[datetime]
  • status_insert_datetime : datetime

LLMPrompt

LLMPrompt()

Base model to define llm prompts

Attributes

  • role : str
  • task : str
  • behavior_guidelines : str
  • security_guidelines : str

LLMSpecs

LLMSpecs()

Base model to define llm specs

Attributes

  • llm : str
  • temperature : float
  • prompt : LLMPrompt

LocalDataSource

LocalDataSource()

Use this data source if you want to upload a file from your local disk to the ML cube platform cloud.

Attributes

  • file_type : FileType
  • is_folder : bool
  • folder_type : FolderType | None
  • file_path : str

Model

Model()

Base model to define model item

Attributes

  • model_id : str
  • task_id : str
  • name : str
  • version : str
  • metric_name : performance or error metric associated with the model
  • creation_datetime : Optional[datetime]
  • retrain_trigger : Optional[RetrainTrigger]
  • retraining_cost : float
  • llm_specs : Optional[LLMSpecs]

MonitoringQuantityStatus

MonitoringQuantityStatus()

Base model to store the monitoring status of a monitoring quantity (target or metric)

Attributes

  • monitoring_target : MonitoringTarget
  • status : MonitoringStatus
  • monitoring_metric : MonitoringMetric | None
  • segment_id : str | None

MqttNotificationAction

MqttNotificationAction()

Base Model for Mqtt Notification Action

Attributes

  • type : DetectionEventActionType.MQTT_NOTIFICATION
  • connection_string : str
  • topic : str
  • payload : str

MulticlassClassificationTaskCostInfo

MulticlassClassificationTaskCostInfo()

Multiclass classification cost info is expressed in terms of the misclassification costs for each class

MultilabelClassificationTaskCostInfo

MultilabelClassificationTaskCostInfo()

Multilabel classification cost info is expressed in terms of false positive and false negative costs for each class

NumericSegmentRule

NumericSegmentRule()

Rule for a segment over numeric values. It contains a list of ranges that are considered in OR logic to define the rule. See SegmentRule for additional details.

Attributes

  • values : list[SegmentRuleNumericRange]

Methods:

.get_supported_data_types

.get_supported_data_types()

Project

Project()

Project model

Attributes

  • project_id : str
  • name : str

RegressionTaskCostInfo

RegressionTaskCostInfo()

Regression cost info is expressed in two terms: - cost due to overestimation - cost due to underestimation

Fields: currency overestimation_cost underestimation_cost

RemoteDataSource

RemoteDataSource()

A source that identifies where data is stored.

Attributes

  • file_type : FileType
  • is_folder : bool
  • folder_type : FolderType | None
  • credentials_id : The id of the credentials to use to authenticate to the remote data source. If None, the default will be used

Methods:

.get_path

.get_path()

Return the path of the object

.get_source_type

.get_source_type()

Returns raw data source type

ResampledDatasetSuggestion

ResampledDatasetSuggestion()

ResampledDatasetSuggestion base model

Attributes

  • suggestion_id : str
  • suggestion_type : SuggestionType
  • sample_ids : List[str]
  • sample_counts : List[int]

RetrainAction

RetrainAction()

Base Model for Retrain Action

Attributes

  • type : DetectionEventActionType.RETRAIN
  • model_name : str

RetrainTrigger

RetrainTrigger()

Base model to define a retrain trigger

Fields: type credentials_id

RetrainingReport

RetrainingReport()

base model for Retraining Report

Attributes

  • report_id : str
  • suggestion : Suggestion
  • effective_sample_size : float
  • model_metric_name : str
  • performance_upper_bound : float
  • performance_lower_bound : float
  • cost_upper_bound : float
  • cost_lower_bound : float

S3DataSource

S3DataSource()

A source that identifies a file in an S3 bucket.

Attributes

  • file_type : FileType
  • is_folder : bool
  • folder_type : FolderType | None
  • credentials_id : The id of the credentials to use to authenticate to the remote data source. If None, the default will be used
  • object_path : str

Methods:

.get_path

.get_path()

.get_source_type

.get_source_type()

SampleWeightsSuggestion

SampleWeightsSuggestion()

SampleWeightsSuggestion base model

Attributes

  • suggestion_id : str
  • suggestion_type : SuggestionType
  • sample_ids : List[str]
  • sample_weights : List[float]

SecretAWSCredentials

SecretAWSCredentials()

AWS integration credentials, that also include the trust policy that you need to set on the IAM role on AWS.

Attributes

  • credentials_id : str
  • name : str
  • default : bool
  • type : ExternalIntegration
  • role_arn : The ARN of the IAM role that should be assumed
  • trust_policy : The trust policy that should be set on the IAM role on AWS

Segment

Segment()

A Segment is a partition of the data, defined by a set of rules that are applied to the DataSchema. Each rule of the segment is applied in AND, whereas the values of each rule are applied in OR.

Attributes

  • segment_id : str
  • name : str
  • rules : list[SerializeAsAny[NumericSegmentRule | CategoricalSegmentRule]]

SegmentRule

SegmentRule()

A segment is composed by a set of rules that are applied over the fields of the DataSchema. Each rule is applied in AND logic with the other rules, and supports an operator that can either be: the value of the field must be in the list of values. the value of the field must not be in the list of values.

Attributes

  • column_name : str
  • operator : SegmentOperator

Methods:

.get_supported_data_types

.get_supported_data_types()

Get the supported data types for the rule

SegmentRuleNumericRange

SegmentRuleNumericRange()

Numeric range for a single element of values in a NumericSegmentRule

Attributes

  • start_value : float | None
  • end_value : float | None

SegmentedMonitoringFlags

SegmentedMonitoringFlags()

Model containing the monitoring flags of a given segment, identified by its id.

Attributes

  • segment_id : str
  • flags : list[DataBatchMonitoringFlag]

SlackNotificationAction

SlackNotificationAction()

Action that sends a notification to a Slack channel through a webhook that you configure.

Attributes

  • webhook : str
  • channel : str type = DetectionEventActionType.SLACK_NOTIFICATION

SubscriptionPlanInfo

SubscriptionPlanInfo()

Data model for a subscription plan Permission limit set to None means no limit is set Expiration date set to None means no expiration is set Product key data are set only if a product key is associated to the subscription plan

Attributes

  • subscription_id : str
  • type : SubscriptionType
  • max_tasks : int | None
  • max_users : int | None
  • monitoring : bool
  • explainability : bool
  • retraining : bool
  • is_active : bool
  • start_date : date
  • expiration_date : date | None
  • product_key : str | None
  • product_key_status : ProductKeyStatus | None

Suggestion

Suggestion()

Suggestion base model

Attributes

  • suggestion_id : str
  • suggestion_type : SuggestionType
  • sample_ids : List[str]

SuggestionInfo

SuggestionInfo()

SuggestionInfo base model

Attributes

  • id : str
  • executed : bool
  • timestamp : float

TabularData

TabularData()

Tabular data model i.e., a data that can be represented via DataFrame and is stored in formats like: csv, parquet, json

Attributes

  • data_structure : DataStructure = DataStructure.TABULAR
  • source : DataSource

Task

Task()

Task model

Attributes

  • task_id : str
  • name : str
  • type : TaskType
  • cost_info : TaskCostInfoUnion | None = None
  • optional_target : bool
  • monitoring_targets : list[MonitoringTarget]
  • monitoring_metrics : ( None | dict[MonitoringTarget, list[tuple[MonitoringMetric, str | None]]]
  • monitoring_status : list[MonitoringQuantityStatus] ) = None

TaskCostInfo

TaskCostInfo()

Base class for task cost info. It depends on TaskType because classification is different from regression in terms of business costs due to errors

TaskLlmSecReportItem

TaskLlmSecReportItem()

Task LLM security report item model. It contains the most important information of a LLM security report.

Attributes

  • id : str
  • creation_date : datetime
  • name : str
  • status : JobStatus
  • from_datetime : datetime
  • to_datetime : datetime

TaskRagEvalReportItem

TaskRagEvalReportItem()

base model for Rag Evaluation Report

Attributes

  • id : str
  • creation_datetime : datetime
  • name : str
  • status : JobStatus
  • from_datetime : datetime
  • to_datetime : datetime

TaskTopicModelingReportDetails

TaskTopicModelingReportDetails()

Task Topic Modeling Report Details base model

Attributes

  • sample_ids : list[str]
  • topics : list[str]

TaskTopicModelingReportItem

TaskTopicModelingReportItem()

Task Topic Modeling Report Item base model

Attributes

  • id : str
  • creation_datetime : datetime
  • name : str
  • status : JobStatus
  • from_date : datetime
  • to_date : datetime

TeamsNotificationAction

TeamsNotificationAction()

Base Model for Teams Notification Action

Attributes

  • type : DetectionEventActionType.TEAMS_NOTIFICATION
  • webhook : str

TextData

TextData()

Text data model for nlp tasks.

Attributes

  • data_structure : DataStructure = DataStructure.TEXT
  • source : DataSource
  • embedding_source : DataSource | None