This thesis investigated the impact of climate change on the number of flight delays and corresponding flight delay costs by examining departure delays from John F. Kennedy Airport in New York City, USA, from 2013 to 2022 and deriving a model for future delays from 2023 to 2030. This is relevant as the airline industry faces high flight delay costs caused by a changing and more extreme climate. First, machine learning algorithms were trained and evaluated on past weather and flight delay data to find the best model to predict whether a flight is delayed or not and the cost of the delay. The best-performing models, a gradient boosting classifier and a gradient boosting regressor, were then used to make predictions on data of two future climate scenarios. These scenarios represent the upper and lower thresholds of the expected evolution of anthropogenic greenhouse gas emissions and resulting climate change. The outcomes showed no significant change in the number of weather-related flight delays and flight delay costs until 2030 based on the computed Kendall Taus and Spearman Rank Correlations. Additionally, the results identified significant differences in the average delay cost per flight between airlines. It was recommended to regularly repeat this research to spot increasing delay risks as early as possible. This thesis applied the business analytics principles by exploring how the airline industry can use the prediction results to make business decisions. Suggestions were cost reduction measures and increasing the quantity or prices of plane tickets or complementary services.
- Flight delays
- Flight delay cost
- Prediction
- Weather data
- Climate change
- Machine learning
- Mestrado em Análise de Dados para Gestão
Forecasting flight delays with climate data and implications for the airline industry
Wimmer, A. C. (Student). 24 Jan 2024
Student thesis: Master's Thesis