A Novel Approach: Airline Delay Prediction Using Machine Learning

Flight delays often create exasperation in airports when not properly mobilized. Increasing development of machine learning models provoke the researchers and scientists to harness the modern research problems. Due to an increase in customer satisfaction in the air transportation system, there needs a proper decision-making process to mobilize the air-traffic with minimum delay. It is recorded that 19% of United States domestic flights reach their destination with an average delay of 15 minutes. Moreover, the complexity of the air transportation system limits the availability of accurate prediction models. Due to the stochastic nature of delays, this research investigates the qualitative prediction of airline delays to implement necessary changes and provide better customer experience. Collection of historical weather data and operational data during departure and arrival at airports serves as the source for building prediction models. A logistic regression model is used to get the status of the delay which is further contrasted to a decision tree model for evaluating the performance of the delay. The proposed research empirically evaluates the effectiveness of the decision tree algorithm over logistic regression. The results of this simulation indicate the potential delays in major airports including the time, day, weather, etc., and hence the volume of delay shall be minimum based on the constructed model.