Why is entity ID and event ID important in Amazon Fraud Detector?

Question

I'm having difficulty finding information about the `entityId` and `eventId` input variables to [get a prediction](https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/frauddetector.html#FraudDetector.Client.get_event_prediction) from Amazon Fraud Detector.

The description is only "*The unique ID used to identify the event.*" and "*The entity ID. If you do not know the entityId , you can pass unknown , which is areserved string literal.*" Can someone explain why this variable is important? What happens if someone uses the same event ID for two different events? How are entity IDs created?

Accepted Answer

**Entity ID** is "who" performed the event. Depending on the type of data, this could be a username, user ID, company name, etc. If you think about an ML model and trying to predict fraud... you could imagine that if UserA has committed fraud in the past, it is more likely that transactions from UserA in the future are fraudulent. So, providing entity ID can help with accuracy of predictions. Though, it is not a required field and that's why you could just pass in `unknown`.

[Source](https://docs.aws.amazon.com/frauddetector/latest/ug/frauddetector-ml-concepts.html)

**Event ID** is just what it sounds like. It's a specific purchase, credit card transaction, etc. I think that if a customer is using a proper data set, one would not expect two records to have the same Event ID. If the customer's data set does not have a unique event ID, they should create a synthetic ID... either a concatenation of other fields that uniquely represent the row, or just generate a UUID.

Why is entity ID and event ID important in Amazon Fraud Detector?

Relevant content