Detector Pipeline
A 4-step recipe with Amazon Fraud Detector. We can help you get access.
Recipe · 4 steps
Detector Pipeline
with Amazon Fraud Detector · Fraud Detection & Investigation Management
Stand up an event type, detector, and rule, then score a sample event.
- Create or update the event type so the detector and event share a schema.
- Create the detector that will orchestrate rules for the event type.
- Author a DETECTORPL rule on the detector with its expression and outcomes.
- Submit a sample event to the detector for prediction and return the model scores and rule results.
What you'll need to have ready
- eventTypeName: The event type the detector and event share.
- eventVariables: The event variable names that make up the event schema.
- labels: The label names available for this event type.
- entityTypes: The entity type names associated with the event type.
- detectorId: The detector identifier to create and score against.
- detectorDescription: A human readable description of the detector.
- ruleId: The identifier for the rule to create on the detector.
- ruleDescription: A human readable description of the rule.
- expression: The rule expression (e.g. "$model_score > 900").
- language: The rule language, such as DETECTORPL.
- outcomes: The outcome names returned when the rule matches.
- eventId: The unique ID of the event being scored.
For your developer
Arazzo workflow amazon-fraud-detector:detector-pipeline, 4 operations: PutEventType → PutDetector → CreateRule → GetEventPrediction. Specs on APIs.io.