Analysis for Tourists Questionnaire and PPG signal

Abstract

Traditionally, impressions and feedback on tourism experiences have been collected through post-visit surveys. However, capturing tourists' real-time emotional responses during their visits could provide deeper insights into the appeal of destinations and experiences. This study aims to estimate real-time emotional states during tourism using physiological signals, specifically photoplethysmogram (PPG), combined with machine learning techniques. Emotional labels were based on Plutchik’s eight basic emotions. Initial multi-class classification using SVM, Random Forest, and GBDT models yielded insufficient performance. By simplifying the task into binary classification (positive vs. negative emotions), the SVM model achieved an accuracy of 0.870. These results suggest that the proposed method can roughly estimate emotional states from PPG signals, offering a potential tool for real-time emotional analysis in tourism contexts.