Event Detection Optimization Through Stacking Ensemble and BERT Fine-Tuning for Dynamic Pricing of Airline Tickets

Abstract

Dynamic pricing of airline tickets in competitive markets demands innovations that respond to fluctuating market conditions. Public events, such as sporting events, music concerts, and natural disasters, significantly influence pricing strategies. This study aims to optimize airline ticket revenue by improving flight occupancy during such events. Data were collected from Twitter (X) and categorized into eight event types: soccer events, music concerts, volcanic eruptions, earthquakes, riots, floods, motorcycle racing, and non-events. A stacking ensemble method was applied for data labeling, and a fine-tuned BERT model was employed for event detection. The stacking ensemble achieved an accuracy of 0.99, while the fine-tuned BERT model reached an accuracy of 0.94. These results highlight substantial improvements in the accuracy and effectiveness of dynamic pricing strategies. The findings provide a robust solution to dynamic pricing challenges and offer new opportunities to increase revenue through event sentiment analysis, enhancing competitiveness and market flexibility. This research lays the groundwork for developing more adaptive dynamic pricing models by leveraging the combined strengths of stacking ensemble techniques and BERT model fine-tuning to improve overall accuracy.

Keywords

• Dynamic pricing
• event detection
• stacking ensemble
• BERT model fine-tuning