Question 800
A fraud model emits several predictions for the same account every day. Weeks later, verified outcomes arrive, but an evaluation job joins outcomes only on account ID and attaches the same label to all past predictions. Which data-joining contract should be introduced?
Explanation
Correct Answer
Multiple predictions for one account need event-specific correlation and time semantics. A defined maturity window prevents misattributing later outcomes to earlier decisions.
Incorrect Answers
Filling unobserved labels with negatives fabricates ground truth and biases the measured performance.
A model version groups many unrelated predictions and cannot identify which event the observed outcome evaluates.
Account identity alone does not distinguish separate decisions or the event that each outcome should label.