Understanding tourism demand: A tree-based Analysis of territorial drivers

We investigate the territorial drivers of tourism arrivals to support more effective destination planning. The analysis uses a Multivariate Regression Tree (MRT), an interpretable method that captures non-linearities and interactions while identifying homogeneous groups of destinations. The study focuses on 307 municipalities in Sardinia (Italy), using tourist arrivals by nationality as response variables and geographic, demographic, and accessibility factors as predictors. Results show that accommodation density is the main determinant of tourism attractiveness and reveal a marked coastal–inland divide driven by the interaction between supply and accessibility. The study highlights the value of tree-based methods for analysing tourism demand in heterogeneous contexts.