A novel approach to online review analysis: integrating theory of planned behavior and machine learning techniques

Purpose This study aims to examine the applicability of the Theory of Planned Behavior (TPB) using online consumer reviews to better understand and predict actual revisit behavior in the hospitality industry. Using unstructured text data and machine learning techniques, it aims to bridge the gap between behavioral intentions and actual behavior. Design/methodology/approach This study constructs the TPB components (attitudes, social norms and perceived control) from textual data, and analyzes their impacts on hotel customers’ behavioral intentions and actual revisit behavior, using Generalized Structural Equation Modeling. Neural networks are applied to assess the predictive accuracy of TPB constructs in forecasting actual revisit behavior. Findings The findings validate the presence of TPB constructs within text data and demonstrate their significant influence on revisit intentions and actual behavior. Mediation analysis confirms the role of revisit intention in linking TPB constructs to actual behavior. Neural networks improve prediction accuracy, highlighting the potential of combining theories and advanced machine learning techniques to analyze consumer behavior. Practical implications This study provides practical insights for hospitality practitioners, emphasizing how managers can utilize consumer reviews to identify psychological drivers of customer satisfaction and behaviors and inform data-driven decision-making strategies. Originality/value This research offers a novel methodological framework by integrating TPB with machine learning techniques to analyze unstructured text data. Unlike previous studies, it goes beyond examining intentions by incorporating actual behavior, contributing to the theoretical development of TPB and practical advancements in consumer behavior analysis.