Decision thresholds
The cut-off score at which a model’s output triggers an action.
Many models output a probability; the threshold turns it into a decision, for example flagging transactions above 0.8. Moving it trades false positives against false negatives. Thresholds should be approved, documented and change-controlled.
On the exam
A threshold change can change outcomes as much as retraining does, so it needs change management.
Related terms
- Confusion matrix (false positives and false negatives)A table of a model’s correct and incorrect predictions.
- Change management for AIControlling changes to models, data, prompts, thresholds and settings.
AAIA Test yourself on this with timed AAIA practice questions in the AAIA Exam Simulation Hub, or read the free AAIA exam guide.
Test yourself with today’s question of the day.
