Predictive Modelling
AI-Powered Flight Disruption Prediction
Project
Challenge
Flight delays, cancellations, and diversions disrupt airline operations, increase costs, and frustrate passengers. Airlines need a predictive system to anticipate disruptions in advance, allowing them to take proactive measures to improve efficiency and customer experience.
Goal
To develop an AI-driven flight status prediction system that classifies flights as on-time, delayed, cancelled, or diverted using historical flight data. By analysing key factors such as scheduled departure time, flight distance, and elapsed time, this system provides real-time insights to mitigate disruptions before they happen.
Result
- 98% accurate predictions using a tree-based Random Forest model. - Key risk factors identified: Scheduled elapsed time, flight distance, and departure time. - High-risk periods detected: January sees 97% of delays, February has peak diversions. - Disruption-prone routes identified: LAS, PHX, and MDW show higher delay rates. - Airlines can now anticipate disruptions and proactively manage operations to minimise delays and improve passenger experience.