Airfare prediction: Leveraging market data for better decision-making

Abstract

Airline revenue management is crucial for airlines to maintain their competitive position in the market. Revenue management addresses two main concerns in airline planning processes, pricing and seat inventory management, to balance supply and demand. Pricing or determination of airfare is a complex decision-making process influenced by factors including distance, number of passengers, market share, competition, and route-related characteristics. However, it is a central element as it impacts revenue generation, market positioning, demand management, cost recovery, and customer relationships. This study investigates the machine learning perspective on predicting airline market-level airfares and examines the determinants of airfare. In this regard, exploiting the publicly available data from the US Department of Transportation Bureau of Transportation Statistics, several supervised machine learning algorithms are tested and compared to obtain the most effective prediction for the given dataset. The Random Forest model outperformed the other models, with $R_{\mathrm{adj}}^2$ and RMSE scores of 0.998 and 1.811, respectively. An ad hoc feature importance analysis is also performed to gain further insight into the determinants of market-level airfares. The findings emphasize the importance of operational costs and pricing strategies in airfare prices.