Uçak içi satış platformlarında ikram tahminleme

In the competitive landscape of commercial aviation, efficient resource management is pivotal for both environmental sustainability and economic performance. This thesis introduces a novel application of AI models in predicting food and beverage consumption for buy-on-board services across different flight legs. By analyzing historical sales data, the AI model forecasts total consumption, enabling airlines to optimize provisioning and reduce waste. This is critical, as excess inventory not only leads to increased wastage of perishable goods but also contributes to higher fuel consumption due to additional weight. This research not only advances the application of AI in aviation but also serves as a blueprint for enhancing onboard service logistics and sustainability practices. This research focuses on fresh food prediction in BoB system. Fresh Food sales data is used for training and development purposes. Two models are selected for this approach; one is XGBoost and other one is neural networks. These models were selected for their differing strengths in handling complex, nonlinear data relationships. For both models hyperparameter tunings applied and the results evaluated for best values and overfitting conditions. The comparative analysis focused on evaluating their predictive accuracy, robustness, and efficiency in operational environments. Results indicate that while both models enhance forecasting precision over traditional methods, each offers unique advantages that can be leveraged depending on specific operational needs.