Road-Based Tourism Mobility in Italy: Predictive Modelling and Behavioural Profiling through Explainable AI

Road-based tourism represents one of the most widespread forms of domestic travel in Italy. This study investigates the determinants of road-based tourism by combining traditional statistical methods with machine-learning techniques and explainable artificial intelligence (XAI). The analysis is based on microdata from the Italian National Institute of Statistics (ISTAT) Trips and Holidays survey covering the period 2015–2023. Four predictive models were developed and compared: Logistic Regression, Classification Tree, Random Forest, and XGBoost. Model performance was evaluated using multiple classification metrics, including accuracy, precision, recall, F1-score, and ROC-AUC. Among the examined approaches, XGBoost achieved the highest predictive performance. To move beyond prediction and provide behavioural insights, SHAP (SHapley Additive exPlanations) values were employed to interpret both global and local model behaviour. The results reveal that road-based tourism is strongly associated with short-duration trips, leisure travel, summer travel periods, and accommodation types offering greater flexibility, such as second homes, camping facilities, and rental accommodation. Regional differences also emerge, suggesting that the propensity to undertake road-based trips varies substantially across Italy. Furthermore, the analysis highlights the importance of considering interactions between travel characteristics, accommodation choices, and socio-demographic factors when studying mobility behaviour. By integrating predictive modelling with explainable AI, this study contributes to a deeper understanding of domestic travel behaviour and demonstrates how machine-learning methods can complement traditional statistical approaches in tourism research. The findings may support policymakers and tourism stakeholders in designing more targeted strategies for mobility planning, destination management, and tourism development.