Comparison of Different Machine Learning Approaches, for Predicting Aircraft Departure Delays, Due to the Circumstances of the Defrosting Process

One of the standard procedures carried out at airports during winter months is the de-icing and anti-icing procedure for aircraft. Circumstances accompanying this procedure such as aircraft type, amount of fluid used, external temperature, wind speed and dew point, directly affect the duration of the procedure and lead to delays as they are conducted immediately before takeoff. A useful tool that could efficiently predict the length of departure delays based on current circumstances, alert airport staff, and indicate the need for additional measures to reduce undesirable delays would be valuable. In this paper, specific data collected at “Konstantin Veliki” Airport in Niš, for the period from November 2020 to March 2024, were processed using two of the most common machine learning methods: regression analysis and classification. The results were compared, and the method that exhibited superior statistical parameters and more accurate prediction was proposed for practical use.