Decision support for airfare forecasting using an evolutionary approach: Evidence on search engine data from Touristic agency

Abstract

This study addresses the complex issue of airfare forecasting pricing that influences decisions in the Airline Industry. Our approach involves an evolutionary methodology, combining statistical analysis and symbolic regression through grammar-guided genetic programming models. This innovative method aids internet users/clients in determining optimal periods to purchase fixed-date flight tickets. Data analysis identifies key relations, guiding effective data pre-processing. Results, based on novel data from the search engine of a large European Travel Online agency, cover one and a half years and millions of airfare search cases for seven travel destinations. Benchmarking against last year's price variation, our forecasting accuracy ranges between 80 and 90 % for the tested travel destinations.