AI-powered hotel performance prediction: a multimodal data approach
Ningqiao Li, Minwoo Lee, Agnes DeFranco, Jae Kyeong Kim, Xinzhe LiPurpose
Hotel-generated social media content functions as key market-facing signals for prospective consumers seeking information to support decision-making, while property-level information serves as structural signals. Despite their growing relevance, limited research has examined the roles of these signals in forecasting hotel performance. This study aims to address this gap. In particular, this study investigates the joint effect of multimodal data (i.e. social media visual and textual content alongside property-level numerical information) on hotel performance prediction through a comparison of multiple machine learning models.
Design/methodology/approach
To achieve these goals, the present study uses an ML-driven predictive modeling approach by employing multimodal data such as social media photos, captions and financial data for major hotel properties’ business performance prediction.
Findings
Results show that the Random Forest model outperforms others in predicting both hotel occupancy and room revenue. Among all features, image aesthetics emerged as the most influential predictor of occupancy, while total marketing expenditure was the strongest predictor of revenue. These findings extend the application of signaling theory in the hotel performance forecasting context and offer practitioners a practical methodological framework for utilizing multi-modal data and ML to monitor and enhance hotel performance.
Originality/value
The present study contributes to tourism and hospitality literature by analyzing how textual and pictorial cues influence consumer decisions. This study introduces a novel framework for predictive modeling, employing deep learning models to enhance predictive power and stability. By extracting diverse features from social media content and financial data, the study offers deeper insights into consumer decision-making processes. Practically, it helps hotels forecast occupancy rates and total room revenue, aiding in strategic resource allocation in areas such as human resources, marketing expenses, communication and social media advertising. The findings enable managers to identify critical factors for higher performance, allowing timely adjustments to maximize overall revenue. These improvements will enhance the competitiveness and long-term viability of hotel properties.