A Data-driven and Process-aware Approach for Predicting the Turnaround Time of an Aircraft

Aircraft turnaround time prediction is a topic that has gained an increasing at-
tention within airports operations management research during the last years.
It has proven to be an eective mean to dynamically adapt the management of
airport ground operations to react to delays in airplane departure/arrival, e.g.,
to limiting possible cascade eects. In this paper, we introduce a process-aware
and data driven approach to predict turnaround times of an aircraft. With
respect with state of the art approaches, our method combines the advantages
of exploiting machine learning to generate data-driven predictions and the use
of turnaround process model to enhance the comprehension of the generated
predictions for an human decision maker. In particular, taking into account the
structure of the turnaround process allows us to better supporting the involved
human actors in determining at which point of the process the problem is expected and, consequently, which counter actions are possible. Process mining is
used to guarantee the fidelity of the employed turnaround process model with
respect to real-world executions. We developed and tested our approach in a
real case study, i.e., a Dutch airport. The computational experiments show
that the proposed approach delivers high-quality and robust predictions with a
certain transparency in ground operation activities.