Model Version Representation

ModelVersionRepresentation ¶

Represents the metadata of a Vectice model version.

A Model Version Representation shows information about a specific version of a model from the Vectice app. It makes it easier to get and read this information through the API.

Hint

A model version ID starts with 'MDV-XXX'. Retrieve the ID in the Vectice App, then use the ID with the following methods to get the model version: connect.model_version('MDV-XXX') or connect.browse('MDV-XXX') (see Connection page).

Attributes:

Name	Type	Description
`id`	`str`	The unique identifier of the model version.
`project_id`	`str`	The identifier of the project to which the model version belongs.
`name`	`str`	The name of the model version. For model versions it corresponds to the version number.
`status`	`str`	The status of the model version (EXPERIMENTATION, STAGING, PRODUCTION, or RETIRED).
`description`	`str`	The description of the model version.
`technique`	`str`	The technique used by the model version.
`library`	`str`	The library used by the model version.
`iteration_origin`	`str \| None`	The iteration in which this model version was created.
`phase_origin`	`str \| None`	The phase in which this model version was created.
`metrics`	`List[Dict[str, Any]]`	The metrics associated with the model version.
`properties`	`List[Dict[str, Any]]`	The properties associated with the model version.
`model_representation`	`ModelRepresentation`	Holds informations about the source model linked to the model version, where all versions are grouped together.

asdict ¶

asdict()

Transform the ModelVersionRepresentation into a organised dictionary.

Returns:

Type	Description
`Dict[str, Any]`	The object represented as a dictionary

download_attachments ¶

download_attachments(attachments=None, output_dir=None)

Downloads attachments of this version.

Parameters:

Name	Type	Description	Default
`attachments`	`list[str] \| str \| None`	Specific attachments to download, if None then all the version attachments will be downloaded.	`None`
`output_dir`	`str \| None`	Destination for the attachments to be downloaded to. If None then attachments will be downloaded to the current directory.	`None`

Returns:

Type	Description
`None`	None.

get_table ¶

get_table(table)

Extracts the Vectice table in attachment of the model version representation into a DataFrame for better readability.

Returns:

Type	Description
`DataFrame`	A pandas DataFrame representing the table of the model version.

list_attachments ¶

list_attachments()

Lists all the attachments (tables, files, pickle...) of the ModelVersionRepresentation.

Returns:

Type	Description
`list[Attachment]`	The attachments list.

metrics_as_dataframe ¶

metrics_as_dataframe()

Transforms the metrics of the ModelVersionRepresentation into a DataFrame for better readability.

Returns:

Type	Description
`DataFrame`	A pandas DataFrame containing the metrics of the model version.

properties_as_dataframe ¶

properties_as_dataframe()

Transforms the properties of the ModelVersionRepresentation into a DataFrame for better readability.

Returns:

Type	Description
`DataFrame`	A pandas DataFrame containing the properties of the model version.

update ¶

update(
    status=None,
    metrics=None,
    properties=None,
    attachments=None,
)

Update the Model Version from the API.

Parameters:

Name	Type	Description	Default
`status`	`str \| None`	The new status of the model. Accepted values are EXPERIMENTATION, STAGING, PRODUCTION and RETIRED.	`None`
`properties`	`dict[str, str \| int] \| list[Property] \| Property \| None`	The new properties of the model.	`None`
`metrics`	`dict[str, int \| float] \| list[Metric] \| Metric \| None`	The new metrics of the model.	`None`
`attachments`	`TAttachment \| None`	The new attachments of the model.	`None`

Returns:

Type	Description
`None`	None