OpenMetadata/openmetadata-docs/content/v1.2.0-SNAPSHOT/main-concepts/metadata-standard/schemas/entity/data/mlmodel.md

title	slug
mlmodel	/main-concepts/metadata-standard/schemas/entity/data/mlmodel

MlModel

This schema defines the Model entity. Machine Learning Models are algorithms trained on data to find patterns or make predictions.

Properties

• id: Unique identifier of an ML Model instance. Refer to ../../type/basic.json#/definitions/uuid.
• name: Name that identifies this ML Model. Refer to ../../type/basic.json#/definitions/entityName.
• fullyQualifiedName: A unique name that identifies an ML Model. Refer to ../../type/basic.json#/definitions/fullyQualifiedEntityName.
• displayName (string): Display Name that identifies this ML Model.
• description: Description of the ML Model, what it is, and how to use it. Refer to ../../type/basic.json#/definitions/markdown.
• algorithm (string): Algorithm used to train the ML Model.
• mlFeatures (array): Features used to train the ML Model. Default: None.
  • Items: Refer to #/definitions/mlFeature.
• mlHyperParameters (array): Hyper Parameters used to train the ML Model. Default: None.
  • Items: Refer to #/definitions/mlHyperParameter.
• target: For supervised ML Models, the value to estimate. Refer to ../../type/basic.json#/definitions/entityName.
• dashboard: Performance Dashboard URL to track metric evolution. Refer to ../../type/entityReference.json.
• mlStore: Location containing the ML Model. It can be a storage layer and/or a container repository. Refer to #/definitions/mlStore.
• server: Endpoint that makes the ML Model available, e.g,. a REST API serving the data or computing predictions. Refer to ../../type/basic.json#/definitions/href.
• href: Link to the resource corresponding to this entity. Refer to ../../type/basic.json#/definitions/href.
• owner: Owner of this ML Model. Refer to ../../type/entityReference.json.
• followers: Followers of this ML Model. Refer to ../../type/entityReferenceList.json#/definitions/entityReferenceList.
• tags (array): Tags for this ML Model. Default: None.
  • Items: Refer to ../../type/tagLabel.json.
• usageSummary: Latest usage information for this ML Model. Refer to ../../type/usageDetails.json. Default: None.
• version: Metadata version of the entity. Refer to ../../type/entityHistory.json#/definitions/entityVersion.
• updatedAt: Last update time corresponding to the new version of the entity in Unix epoch time milliseconds. Refer to ../../type/basic.json#/definitions/timestamp.
• updatedBy (string): User who made the update.
• service: Link to service where this pipeline is hosted in. Refer to ../../type/entityReference.json.
• serviceType: Service type where this pipeline is hosted in. Refer to ../services/mlmodelService.json#/definitions/mlModelServiceType.
• changeDescription: Change that lead to this version of the entity. Refer to ../../type/entityHistory.json#/definitions/changeDescription.
• deleted (boolean): When true indicates the entity has been soft deleted. Default: False.
• extension: Entity extension data with custom attributes added to the entity. Refer to ../../type/basic.json#/definitions/entityExtension.
• domain: Domain the MLModel belongs to. When not set, the MLModel inherits the domain from the ML Model Service it belongs to. Refer to ../../type/entityReference.json.

Definitions

• featureType (string): This enum defines the type of data stored in a ML Feature. Must be one of: ['numerical', 'categorical'].
• featureSourceDataType (string): This enum defines the type of data of a ML Feature source. Must be one of: ['integer', 'number', 'string', 'array', 'date', 'timestamp', 'object', 'boolean'].
• featureSource (object): This schema defines the sources of a ML Feature. Cannot contain additional properties.
  • name: Refer to ../../type/basic.json#/definitions/entityName.
  • dataType: Data type of the source (int, date etc.). Refer to #/definitions/featureSourceDataType.
  • description: Description of the feature source. Refer to ../../type/basic.json#/definitions/markdown.
  • fullyQualifiedName: Refer to ../../type/basic.json#/definitions/fullyQualifiedEntityName.
  • dataSource: Description of the Data Source (e.g., a Table). Refer to ../../type/entityReference.json.
  • tags (array): Tags associated with the feature source. Default: None.
    • Items: Refer to ../../type/tagLabel.json.
• mlFeature (object): This schema defines the type for an ML Feature used in an ML Model. Cannot contain additional properties.
  • name: Refer to ../../type/basic.json#/definitions/entityName.
  • dataType: Data type of the column (numerical vs. categorical). Refer to #/definitions/featureType.
  • description: Description of the ML Feature. Refer to ../../type/basic.json#/definitions/markdown.
  • fullyQualifiedName: Refer to ../../type/basic.json#/definitions/fullyQualifiedEntityName.
  • featureSources (array): Columns used to create the ML Feature. Default: None.
    • Items: Refer to #/definitions/featureSource.
  • featureAlgorithm (string): Description of the algorithm used to compute the feature, e.g., PCA, bucketing...
  • tags (array): Tags associated with the feature. Default: None.
    • Items: Refer to ../../type/tagLabel.json.
• mlHyperParameter (object): This schema defines the type for an ML HyperParameter used in an ML Model. Cannot contain additional properties.
  • name (string): Hyper parameter name.
  • value (string): Hyper parameter value.
  • description: Description of the Hyper Parameter. Refer to ../../type/basic.json#/definitions/markdown.
• mlStore (object): Location containing the ML Model. It can be a storage layer and/or a container repository. Cannot contain additional properties.
  • storage: Storage Layer containing the ML Model data. Refer to ../../type/basic.json#/definitions/href.
  • imageRepository: Container Repository with the ML Model image. Refer to ../../type/basic.json#/definitions/href.

Documentation file automatically generated at 2023-07-07 05:50:35.981927.