Airline revenue management with demand forecast updating: a case study of single-leg data

This study investigates the impact of demand forecast updating on airline revenue management, using two years of single-leg booking data. We develop an integrated model that combines clustering, classification, and demand unconstraining to improve forecasting performance. Three machine learning algorithms—Lasso, LightGBM, and Multi-layer Perceptron—are evaluated for their predictive accuracy. While early updates may slightly reduce forecast precision, the resulting revenue improvements from later updates remain consistently significant across all flights and models. These findings underscore the value of advanced forecasting techniques and continuous updates in optimizing airline revenue in competitive markets.