ARGE-SCALE AIRLINE TICKET PRICE PREDICTION USING ENSEMBLE MACHINE LEARNING ALGORITHMS

Airline ticket price prediction represents a highly complex and dynamic challenge, primarily due to the multifactorial and time-sensitive nature of airline pricing strategies. Accurate forecasting of ticket prices holds substantial value for both consumers, by enabling optimal purchase decisions, and airline companies, by supporting data-driven revenue management and dynamic pricing. In this study, we conduct a comprehensive analysis of a large-scale flight booking dataset obtained from the “Ease My Trip” platform, encompassing over 300,000 records of flight options between major Indian metropolitan cities. A suite of advanced machine learning algorithms, including Linear Regression, CatBoost, LightGBM, Random Forest, and XGBoost, was implemented to model and predict ticket prices. A comparative evaluation of these models reveals that ensemble and boosting algorithms, particularly XGBoost and Random Forest, deliver superior predictive performance, with XGBoost achieving an R² of 0.98 and a mean absolute error (MAE) of $2,035.51. These findings underscore the critical importance of employing robust machine learning techniques and incorporating a diverse set of features for reliable airline ticket price prediction. The results offer valuable insights for both passengers seeking cost-effective travel and airline stakeholders aiming to optimise revenue management strategies.