Forecasting flight delays using machine learning

Abstract

This article considers machine learning and its utilization in the domain of air transportation. The first part of this research aims to define machine learning, describe its historical development and helps delineate machine learning in a broader framework of other data analysis approaches such as artificial intelligence, deep learning and data science. In the second part, the research is meant to explain what machine learning is, tell more about the types of machine learning and how it is used in different scenarios in the aviation industry. This part of the thesis discusses certain areas and real examples of how machine learning is used by multiple companies (aircraft manufacturers) as well as examines the available conclusions of researches already undertaken in the area and determines their connection to the current one. Finally, the practical part of the thesis uses the collected real-time data about departures from two American airports, analyses it with the help of statistical Python-based tools, describes the developed machine-learning algorithm to predict delays, runs experiments on data and discusses results.