A Dynamic Predictive Likelihood Framework for Anticipatory Adaptive Nested Optimization in Airline Seat Reservation Control

This study develops a novel dynamic framework for anticipatory and adaptive airline seat inventory control under stochastic customer demand. The proposed approach integrates predictive likelihood estimation, nested protection-level optimization, and adaptive decision updating within a unified mathematical structure. Demand arrivals are modeled over a discrete booking horizon divided into multiple decision epochs, allowing continuous revision of protection levels and booking limits as new information becomes available. Unlike classical static approaches such as Littlewood's rule and EMSR heuristics, the proposed model combines global non-nested optimization with local nested protection-level adjustment based on predictive likelihood functions of customer demand. The framework enables real-time updating of allocation decisions while maintaining consistency with capacity constraints and revenue dominance conditions. In addition, a statistical procedure based on Weibull modeling of reservation times is introduced to detect structural changes in customer demand or competitive pricing strategies through minimization of misrecognition probability. Numerical simulations demonstrate that the proposed dynamic approach improves expected revenue relative to non-nested allocation and heuristic EMSR methods under representative parameter settings. The results suggest that predictive likelihood-based adaptive optimization provides a rigorous and computationally tractable alternative to heuristic seat allocation rules.