Custom scoring metrics

class datarobot.models.custom_scoring_metrics.CustomScoringMetric

A versioned custom scoring metric container.

Variables:

  • id (str) – Unique identifier.

  • name (str) – Display name of the metric.

  • target_type (str) – The target type this metric applies to (e.g. Binary, Regression, Multiclass).

  • user_id (str or None) – ID of the user who created the metric.

  • organization_id (str or None) – ID of the owning organization.

  • description (str or None) – Optional description.

  • latest_version_num (int) – The highest version number that has been created.

  • is_archived (bool) – Whether the metric has been soft-deleted.

  • created_at (str or None) – ISO-8601 creation timestamp.

  • updated_at (str or None) – ISO-8601 last-updated timestamp.

  • tags (list of dict) – Tag entries, each {"name": str, "value": str} (additional keys allowed).

classmethod list(offset=0, limit=20, target_type=None, search=None, is_archived=False, tags=None)

List custom scoring metrics.

Parameters:

  • offset (int) – Number of records to skip.

  • limit (int) – Maximum number of records to return.

  • target_type (str, optional) – Filter by target type (e.g. “Binary”, “Regression”, “Multiclass”).

  • search (str, optional) – Filter by name sub-string.

  • is_archived (bool) – When True, return only archived metrics. When False (default), non-archived metrics are returned and isArchived is omitted from the list request.

  • tags (list of dict, optional) – Tag entries shaped as {"name": ..., "value": ...}.

Return type:

list[CustomScoringMetric]

classmethod create(name, target_type, description=None, tags=None)

Create a new custom scoring metric container.

Parameters:

  • name (str) – Display name for the metric.

  • target_type (str) – Target type this metric applies to.

  • description (str, optional) – Optional description.

  • tags (list of dict, optional) – Tag entries shaped as {"name": ..., "value": ...}.

Return type:

CustomScoringMetric

classmethod get(metric_id)

Retrieve a custom scoring metric by ID.

Parameters:

metric_id (str) – The ID of the metric to retrieve.

Return type:

CustomScoringMetric

update(name=None, description=None, tags=None)

Update the name, description, or tags of this metric.

Parameters:

  • name (str, optional) – New display name.

  • description (str, optional) – New description.

  • tags (list of dict, optional) – Replacement tags, each {"name": ..., "value": ...}.

Returns:

This instance, updated in place.

Return type:

CustomScoringMetric

delete()

Archive (soft-delete) this metric.

Return type:

None

class datarobot.models.custom_scoring_metrics.CustomScoringMetricVersion

A single version of a custom scoring metric.

Versions are created by uploading a metadata.yml and custom_metrics.py file pair. Each new version receives the next sequential version number.

Variables:

  • id (str) – Unique identifier of this version.

  • custom_scoring_metric_registry_id (str) – ID of the parent metric container.

  • version_num (int or None) – Sequential version number assigned on creation.

  • files_catalog_id (str or None) – ID of the files catalog entry storing the uploaded files.

  • files_catalog_version_id (str or None) – ID of the specific catalog version.

  • metrics (list) – Metric definitions parsed from the uploaded metadata.

  • stage (str or None) – Lifecycle stage of this version.

  • description (str or None) – Optional description provided at upload time.

  • validation_status (str or None) – Status of server-side validation of the uploaded files.

  • user_id (str or None) – ID of the user who created this version.

classmethod list(metric_id, offset=0, limit=20)

List all versions for a given custom scoring metric.

Parameters:

  • metric_id (str) – ID of the parent metric container.

  • offset (int) – Number of records to skip.

  • limit (int) – Maximum number of records to return.

Return type:

list[CustomScoringMetricVersion]

classmethod create(metric_id, metadata_file_path, code_file_path, description=None)

Upload a new version of a custom scoring metric.

Sends metadata.yml and custom_metrics.py as multipart form data.

Parameters:

  • metric_id (str) – ID of the parent metric container.

  • metadata_file_path (str) – Local path to the metadata.yml file.

  • code_file_path (str) – Local path to the custom_metrics.py file.

  • description (str, optional) – Optional description for this version.

Return type:

CustomScoringMetricVersion

classmethod get(metric_id, version_id)

Retrieve a specific version of a custom scoring metric.

Parameters:

  • metric_id (str) – ID of the parent metric container.

  • version_id (str) – ID of the version to retrieve.

Return type:

CustomScoringMetricVersion

delete()

Delete this version.

Return type:

None

classmethod get_file_contents(registry_id, registry_version_id)

Retrieve the raw metadata and code file contents for a version.

Parameters:

  • registry_id (str) – ID of the parent custom scoring metric registry.

  • registry_version_id (str) – ID of the registry version whose files should be retrieved.

Return type:

CustomScoringMetricVersionFileContents

get_files()

Retrieve the raw metadata and code file contents for this version.

Return type:

CustomScoringMetricVersionFileContents

class datarobot.models.custom_scoring_metrics.CustomScoringMetricVersionFileContents

Raw file contents (metadata + code) backing a custom scoring metric version.

Variables:

  • metadata_file_name (str) – Stored name of the metadata file (e.g. metadata.yml or metadata.yaml).

  • metadata_file_contents (str) – Raw UTF-8 contents of the metadata file.

  • code_file_name (str) – Stored name of the Python file (e.g. custom_metrics.py).

  • code_file_contents (str) – Raw UTF-8 contents of the custom metrics Python file.