If the Conflict Detection and Resolution use case was focused on an explanation of the AI outcome provided through the interface, the Delay Prediction use case involves an algorithmic explanation.
ARTIMATION developed three different Machine Learning algorithms to provide transparency to the AI decision-making process to predict a delay on a single runway, and therefore optimise delays, take-offs and runway inspections. Improving trust in these kind of tasks is fundamental, but is the certainty of the outcome the only variable influencing the trustworthiness of a delay prediction?
SHAP, LIME and DALEX are the three algorithms that ARTIMATION is testing these days with an online validation.
If you are an Air Traffic Controller and you would like to try the Delay Prediction tool, you can find more specifics and the exercises easily here!
