A machine learning approach for prediction of on-time performance of flights

One of the major business problems that airlines face is the significant costs that are associated with flights being delayed due to natural occurrences and operational shortcomings, which is an expensive affair for the airlines, creating problems in scheduling and operations for the end-users thus causing bad reputation and customer dissatisfaction. In our paper, a two-stage predictive model was developed employing supervised machine learning algorithms for the prediction of flight on-time performance. The first stage of the model performs binary classification to predict the occurrence of flight delays and the second stage does regression to predict the value of the delay in minutes. The dataset used for evaluating the model was obtained from historical data which contains flight schedules and weather data for 5 years. It was observed that, in the classification stage, Gradient Boosting Classifier performed the best and in the regression stage, Extra-Trees Regressor performed the best. The performance of the other algorithms is also extensively documented in the paper. Furthermore, a real-time Decision Support Tool was built using the model which utilizes features that are readily available before the departure of an airplane and can inform passengers and airlines about flight delays in advance, helping them reduce possible monetary losses.