plexus.dashboard.api.models.score_result module

class plexus.dashboard.api.models.score_result.Score

Bases: object

id: str

name: str

Bases: BaseModel

Represents a single classification or scoring result in the Plexus dashboard.

ScoreResult is the core data structure for tracking individual scoring operations, used both for real-time scoring and evaluation. It integrates with the GraphQL API to provide:

Score value tracking with confidence
Metadata storage for debugging and analysis
Links to related entities (items, accounts, scorecards)
Evaluation result tracking
Batch processing support

Common usage patterns: 1. Creating a single score result:

result = ScoreResult.create(
client=client, value=0.95, itemId=”item-123”, accountId=”acc-456”, scorecardId=”card-789”, metadata={“source”: “phone_call”}

)

Batch creation for efficiency:

results = ScoreResult.batch_create(client, [

{
“value”: 0.95, “itemId”: “item-123”, “accountId”: “acc-456”, “scorecardId”: “card-789”

}, {

“value”: 0.82, “itemId”: “item-124”, “accountId”: “acc-456”, “scorecardId”: “card-789”

}

])
Updating with evaluation results:

result.update(
correct=True, evaluationId=”eval-123”, metadata={“human_label”: “Yes”}

)

ScoreResult is commonly used with: - Evaluation for accuracy testing - ScoringJob for batch processing - LangGraphScore for LLM-based classification

__init__(id: str, value: str, itemId: str, accountId: str, scorecardId: str, scoreId: str | None = None, scoreVersionId: str | None = None, score: Dict[str, Any] | None = None, explanation: str | None = None, confidence: float | None = None, metadata: Dict | None = None, trace: Dict | None = None, correct: bool | None = None, cost: Dict | None = None, attachments: List[str] | None = None, scoringJobId: str | None = None, evaluationId: str | None = None, feedbackItemId: str | None = None, code: str | None = '200', type: str | None = 'prediction', status: str | None = 'COMPLETED', createdAt: datetime | None = None, updatedAt: datetime | None = None, client: _BaseAPIClient | None = None)

accountId: str

attachments: List[str] | None = None

classmethod batch_create(client: _BaseAPIClient, items: List[Dict]) → List[ScoreResult]: Create multiple score results in a single API request.

code: str | None = '200'

confidence: float | None = None

correct: bool | None = None

cost: Dict | None = None

classmethod create(client: _BaseAPIClient, value: float, itemId: str, accountId: str, scorecardId: str, scoringJobId: str | None = None, evaluationId: str | None = None, code: str = '200', type: str = 'prediction', status: str = 'COMPLETED', **kwargs) → ScoreResult

Create a new score result.

Args:

client: The API client instance value: Score value (required) itemId: ID of scored item (required) accountId: Account context (required) scorecardId: ID of scorecard used (required) scoringJobId: ID of scoring job (optional) evaluationId: ID of evaluation (optional) code: HTTP response code (defaults to ‘200’ for success) type: Type of score result - “prediction”, “evaluation”, etc. (defaults to ‘prediction’) status: Status of score result (defaults to ‘COMPLETED’) **kwargs: Optional fields:

confidence: float

metadata: dict (will be JSON serialized)

trace: dict (will be JSON serialized)

cost: dict (will be JSON serialized)

correct: bool

attachments: List[str]

Note:

Either scoringJobId or evaluationId should be provided, but not required

createdAt: datetime | None = None

evaluationId: str | None = None

explanation: str | None = None

feedbackItemId: str | None = None

classmethod fields() → str: Return the GraphQL fields to query for this model

classmethod find_by_cache_key(client: _BaseAPIClient, item_id: str, type: str, score_id: str, account_id: str | None = None) → ScoreResult | None

Find the most recent ScoreResult using cache key components.

This method encapsulates the cache lookup logic for finding existing ScoreResults based on the standard cache key: itemId + type + scoreId. It uses the time-based GSI to return the most recent result by updatedAt.

Args:

client: API client for database operations item_id: Item ID (should be resolved DynamoDB ID) type: Type of score result (should be “prediction” or “evaluation”) score_id: Score ID (should be resolved DynamoDB ID) account_id: Optional account ID for additional context/validation

Returns:

Most recent ScoreResult matching the cache key, or None if not found

Example:

# Find cached score result cached_result = ScoreResult.find_by_cache_key(

client=client, item_id=”da270073-83ab-4c43-a1e6-961851c13d92”, type=”prediction”, score_id=”687361f7-44a9-466f-8920-7f9dc351bcd2”

)

classmethod from_dict(data: Dict[str, Any], client: _BaseAPIClient) → ScoreResult: Create a ScoreResult instance from API response data.

classmethod get_by_id(score_result_id: str, client: _BaseAPIClient) → ScoreResult | None

Get a ScoreResult by its ID.

Args:: score_result_id: The ID of the score result to retrieve client: The API client to use
Returns:: ScoreResult object if found, None otherwise

itemId: str

metadata: Dict | None = None

property numeric_value: float | None: Get the numeric value as a float, or None if invalid.

score: Dict[str, Any] | None

scoreId: str | None = None

property scoreName: str | None: Get the score name from the related score object.

scoreVersionId: str | None = None

scorecardId: str

scoringJobId: str | None = None

status: str | None = 'COMPLETED'

trace: Dict | None = None

type: str | None = 'prediction'

update(**kwargs) → ScoreResult: Update this score result with new values.

updatedAt: datetime | None = None

value: str