Forecasting İstanbul Airport’s Passenger Volume: A Comparative Analysis of Time Series and Machine Learning Models

The volume of air transportation has increased dramatically thanks to globalization. In this perspective, accurate passenger demand forecasting gained importance for proactive capacity planning, resource allocation, and operational efficiency in the global aviation industry. This study aims to forecast passenger numbers in İstanbul Airport (IST), which is the biggest airport of Türkiye and one of the busiest airports in Europe. To achieve this, two methods, namely SARIMA, which is one of the traditional time series methods and the Prophet, which is a machine learning method, were utilized in a comparative analysis, and forecasts for passenger numbers were made with both methods. The study dataset consists of 89 months of passenger data of İstanbul Airport (from January 2019 to May 2026) and data used by dividing as a training and a test set. Methods’ performances were evaluated using standard error metrics, MAPE and RMSE. The empirical findings reveal that SARIMA outperformed the Prophet model in terms of errors. Although the Prophet performed similarly to SARIMA in regular periods, in irregular periods SARIMA performed much better. For the future, both methods are expecting the continuation of the growth trend of İstanbul Airport’s passenger volume.