Flight Cancellation Forecasting Based on Machine Learning Approach

The topic of this work: Flight cancellation forecasting based on machine learning approach. The purpose of this work is to identify the key factors of delay and cancellation of the flight using data analysis and the construction of predictive models. Two main groups of factors were taken into account. The first is flight characteristics, such as the airline, departure and destination airports, the distance between them, etc. The second is the weather at the airport of arrival and departure, which includes temperature, the presence of clouds and their type, pressure, precipitation, etc. The object of study - the market of passenger traffic. The subject of the research is the success of the flight, which includes its status (completed, canceled, delayed), as well as the duration of its delay. Research methods. To implement the practical part of the research, Python 3.8 was chosen. The Python programming language has recently been increasingly used for data analysis, both in science and in the commercial field. This is facilitated by the simplicity of the language, as well as a wide variety of open libraries. For the analysis, data on all domestic flights of the US from the BTS for 2017 and 2018, as well as hourly weather observations from meteorological stations located at the airports, from the site of the NOAA for the same period were used. As a result of the work, an application for passengers was created to predict flight cancellations and delays. It will be useful for those people who buy tickets for connecting flights, and it is important for them that all flights be carried out on time, because otherwise they may be late for the next plane. It will also be useful for busy people with tight work schedules, who cannot afford possible delays on the way and who need to know the risks in advance in order to anticipate them. Thus, it contributes to reducing business costs and increasing the income of the tourist market.