This is a tool for predicting delay propagation in a network of airports. Goal of this is to better
predict and explain
the European
and International airspace such that the unnecessary and preventable delays can be avoided. The
representation of the
network of airports
was done with a Graph Neural Network. In order to account for the spatial and time depended relations of
each node in
the network a
Spatial Temporal Graph Neural Network was used.
The project created as part of the capstone project of the
Engineering with AI minor
at TU Delft.
Learn more about the project on Github.
Press play on the bottom right to show the network over time. Each dot/node represents an airport in Europe's top 50 airports, click on a dot for more info. The size of each dot indicates the magnitude of delays whereas the color of the dot indicates the error of the prediction. Click on "show legend" in the top right corner to understand the colors better.
Constantinos Aristodemou,
Vlad Buzetelu,
Tristan Dijkstra,
Tim Hogenelst,
Niels Prins,
Benjamin Slijper,
Theodor Falat.