Enhancing Flight Delay Prediction: A Deep Learning Approach for Weather-Driven Disruptions

This study proposes a hybrid deep learning model combining Multi-layer LSTM, Attention, and MLP to improve weather-induced flight delay prediction. Using filtered U.S. aviation and weather datasets, the model captures both short- and long-term temporal patterns while dynamically prioritising influential features. Experimental results show superior performance compared with traditional machine learning and standard deep learning models, achieving over 96% accuracy. The approach supports more reliable delay forecasting, enabling airlines and airports to enhance operational planning, resource allocation, and disruption management.