Abstract
<jats:p>In the dynamic airline market, one of the main challenges for companies is the high volatility of ticket prices, which complicates planning and decision-making. This paper investigates approaches and technologies for fore-casting airline ticket prices under multiple parameters. It describes an integrated information system designed for short-term price prediction on popular routes. The data‑collection module aggregates historical sales, demand indicators, seasonal factors, and airline operating parameters, storing them in a PostgreSQL database. Four forecasting approaches are examined at the analytics layer-linear regression, gradient boosting, Prophet, and a Long Short-Term Memory (LSTM) recurrent network. A comparative experiment on five major routes shows that LSTM achieves the best performance, with average errors of RMSE 0.065-0.153 and MAE 0.049-0.117, while remaining robust to outliers. Architecturally, the system is divided into a machine‑learning module (Python + VS Code), a data repository (PostgreSQL), and a visualization layer (Power BI); the deployment diagram illustrates component interaction via a REST API. The user interface offers an interactive price chart and a calendar of minimum fares for the next five days. The reported results confirm the suitability of LSTM for dynamic pricing and demonstrate the scalability of the methodology to finance, energy, and logistics domains where highly-volatile time-series forecasting is required. The conclusion outlines directions for further research and ways to improve forecasting technologies in order to enhance the accuracy and resilience of decision-making in the industry.</jats:p>