Application of Machine Learning for Aviation Safety Risk Metric

Abstract

A risk metric is one of the key tools to monitor the safety performance of complex systems. This paper explains the methodologies employed to develop a new comprehensive risk metric based on Machine Learning to measure the safety performance of the National Airspace System (NAS) in the airport surface environment. Unlike existing metrics that focus primarily on incidents and close calls, this metric incorporates all relevant surface incidents and accidents and assigns an appropriate severity measure proportional to their consequences and in relation to a potential worst outcome of an accident.