Abstract

Airline performance and revenue are closely tied to operational reliability and proactive decision-making. Traditional flight-delay management and fare adjustment systems often operate in isolation, leading to fragmented responses and suboptimal financial outcomes. This paper presents FlightVision Pro, an intelligent framework that unifies flight-delay prediction and dynamic fare optimization through a hybrid ensemble learning approach. The model combines Random Forest, XGBoost, and Neural Network classifiers to forecast both delay occurrence and duration with high accuracy while maintaining low inference latency. These predictions feed into a pricing optimizer that adjusts fares within a ±20 percent margin based on real-time demand and operational risk. The framework is supported by a role-based, offline-capable dashboard built on lightweight components such as SQLite and local model caching, ensuring 48-hour autonomy in limited-connectivity environments. Performance evaluation demonstrates improved predictive reliability (≥85 percent accuracy, MAE ≤ 15 minutes) and revenue gains of 5–15 percent compared with baseline systems. FlightVision Pro thereby establishes a unified operational intelligence layer connecting airline control centers, revenue management teams, and airport coordinators through a single, explainable, and resilient analytics platform.