Deep learning to segment first-time and repeat travelers: Analyzing GPS data

In the competitive tourism landscape, effective market segmentation is essential for targeting high-value segments such as repeat visitors. However, limited research has leveraged en-route trip behaviors to distinguish first-time and repeat travelers. This study proposes travel mobility patterns as a novel explainable variable for data-driven a priori segmentation and evaluates various deep models for classifying predefined traveler types using GPS trajectory data from Incheon. Results indicate that convolutional neural networks outperform alternatives, achieving an average accuracy of 84.53 %, which highlights the promise of deep learning for behavioral segmentation using mobility big data. The findings advance the literature on data-driven a priori segmentation and offer actionable insights for destination marketers to monitor real-time market dynamics and develop tailored strategies.