Airport passenger number prediction using artificial neural network and comparative analysis with some forecast methods

With the increasing competitive environment, it is seen that factors such as time, cost and comfort are given importance in all areas of life, especially in the transportation sectors. The reason for this is that people's desire for less costly, faster and more comfortable transportation increases with the difficulty of living conditions, the increase in population and traffic density. In this direction, it is very important that these sectors should accurately predict the number of passengers for the following months and years in order to increase the demand for a fast and comfortable transportation with less cost. With accurate passenger number estimations; time, energy consumption, customer satisfaction and cost factors can be managed more accurately and practical and strategic plans on issues such as capacity, facility usage, personnel needs, tourism and economy can be provided. By this way investments can be made efficiently and the expected flight prices would decrease accordingly. With lower prices, the fast and comfortable transportation, the demand to these companies will increase and this would help the companies to have prestige. Within the scope of the thesis, Istanbul, Ankara and Izmir are selected from metropolitan cities where there is high transportation, and it is aimed to make accurate estimates of the number of passengers for the airports in these cities. In this direction, an artificial neural network is proposed for the estimating passenger number and this method is compared with other methods. In the application phase; traditional methods, methods that can predict the future, and also seasonality analysis of these methods has been investigated. In order to provide more realistic comparison, traditional methods and methods that can predict the future are evaluated in two separate categories. The mean absolute percentage error value is checked while making the evaluation. As a result of the evaluation, it observed that the artificial neural network is the method that gives the best result and the future passenger number with proposed approach are estimated for years 2024 to 2030.