Modern Practice of Machine Learning in the Aviation Transport Industry

Abstract

The fast-paced life of the present produces huge amounts of data every minute [1]. Timetables, scheduled aircraft maintenance, consumables, loyalty programs and personalized offers for passengers are data that can increase profits and reduce costs [2]. However, analysts are interested in information hidden in the data, which, without proper storage and processing, will remain just a lot of lines and will not be able to bring the expected benefits [3].Machine learning has made a big breakthrough in knowledge processing [4]. Most of the routine operations are algorithmized and transferred to machines; a person, as a rule, acts as a teacher and controller. The indisputable advantage of machining is speed and the minimum probability of making mistakes. For example, in 2019, the average cost of an aircraft delay for a US passenger airline was $ 74.25 per minute.Machine learning is used in the field of airline revenue management: personalized discounts, optimal routes, dynamic formation of air ticket prices, and the like [5]. Correctly laid routes are a very important factor in the success of an airline. A striking example is the presence of direct flights between Krakow (Poland) and Chicago (Illinois, USA), which, at first glance, does not make sense. However, it is enough to analyze the ethnic composition of Chicago to find that 1,500,000 Poles live in the diaspora. Another important factor is dynamic pricing for air tickets. This is not a trivial task, since the airline wants to sell at a higher price, and the passenger wants to buy at a lower price. After all, the price of an air ticket is influenced by many factors: the time of booking, the day of the month, the day of the week, the availability of events at the destination, the cost of fuel, etc. All this data is difficult to process manually, but a properly designed system can handle it easily. Therefore, each airline has its own closed dynamic pricing system, and scientists develop frameworks and train neural networks to accurately predict the cost of air tickets.The development of the aviation industry leads to an increase in the number of flights, which creates more carbon dioxide emissions. In 2018, all commercial air travel caused 2.4% of global CO2 emissions . Although this value may seem insignificant, in 2013-2018 the amount of emissions increased by 32%. Machine

PERSPECTIVES AND PROBLEMS OF BIG DATA ANALYSIS & ANALYTICS FOR EFFECTIVE MARKETING OF TOURISM INDUSTRY

“…All this highly structured, semi-structured and unstructured data is stored in many repositories, often even outside the organization. Companies can have access to vast amounts of their own data and don't need tools that can establish relationships between these data and draw important conclusions based on them [8]. Traditional methods of analysis and analytics cannot be used for huge volumes of constantly growing and updated data, which ultimately paves the way for Big Data technologies [9].…”

Ефективне Застосування Класичних Алгоритмів Машинного Навчання При Прийнятті Адаптивних Управлінських Рішень

“…Алгоритми машинного навчання перетворюють набір даних в модель. Який алгоритм працює найкраще (контрольований, неконтрольований, класифікація, регресія тощо), залежить від типу розв'язуваної задачі, доступних обчислювальних ресурсів і характеру даних [7][8][9].…”