摘要: This paper presents a machine learning technique to predict delays and, based on this forecast, plan flight operations and advisories. Neural networks are also widely researched for predicting flight delays, but these deep learnin... 展开 This paper presents a machine learning technique to predict delays and, based on this forecast, plan flight operations and advisories. Neural networks are also widely researched for predicting flight delays, but these deep learning techniques are mathematically complex. This research adopts Gaussian processes as a supervised learning technique to predict flight delays. It is a data-driven machine learning technique that trains a model with prior knowledge, i.e., mean and covariance functions associated with known data. To predict delays, the model is trained with features like flight day, day of the week, month, departure time, etc. Flights from different routes are considered for prediction, and the efficacy of the presented learning technique is studied by comparison with actual delays of the flights. As it is impossible to predict delay time accurately, we instead predict the delays showing a 95% confidence interval. 收起