Intelligent Fare Dynamics Modeling Using Machine Learning for Adaptive Airline Pricing Forecasts

Abstract

This research addressed the problem of predicting flight ticket prices by analyzing various factors that influence fare variations. Key attributes such as departure time, destination, flight duration, and seasonal demand were considered, as these parameters significantly affect pricing. Flight fares are highly dynamic and tend to fluctuate based on travel schedules, route characteristics, and peak periods such as holidays and vacations, making it important for users to understand pricing trends before booking. In this study, three different datasets were analyzed to extract meaningful insights into fare patterns, and the identified features were applied to seven different Machine Learning (ML) models to compare their predictive performance. The main objective was to identify the most influential factors and develop an effective prediction system. Among the applied techniques, Linear Regression, Decision Tree Regression, and Random Forest Regression were used as core models to enhance accuracy. Linear Regression provided a simple baseline by modeling linear relationships, Decision Tree Regression captured non-linear patterns through hierarchical data splitting, and Random Forest Regression improved robustness by combining multiple decision trees to reduce overfitting. By integrating these approaches, the system effectively captured both simple and complex relationships within the data, resulting in more reliable price predictions. This model can assist users in making informed booking decisions while also providing valuable insights into dynamic pricing behavior in the aviation domain.