A Data-Driven Approach to Air Traffic Delay Prediction and Sentiment Evaluation

Flight delays pose significant challenges to passengers and aviation stakeholders alike, causing not only inconvenience but also substantial economic losses. In this study, we present a machine learning-based approach to predict air traffic delays using real-world flight, airline, and airport datasets. We evaluate multiple regression models including Linear Regression, Lasso, Ridge, Decision Tree, and Random Forest through cross validation, achieving high predictive accuracy, particularly when departure delay is used as a feature. To complement this quantitative analysis, we conduct sentiment analysis on over 1,000 tweets related to flight delays, offering insights into public perception and emotional response. Our results indicate a strong correlation between departure and arrival delays (r 0.94) and highlight the reputational risks airlines face due to negative passenger experiences. The combined methodology demonstrates how predictive analytics and sentiment mining can be leveraged to mitigate the operational and perceptual impact of flight delays.