Andrea Venegoni

Internet searches and query data have become favorable sources for better forecasts of tourist volumes (Pan et al., 2012; Yang et al., 2014; Yang et al., 2015): search engines are used to plan routes, search for hotels, attractions, travel guides and reviews (Fesenmaier et al., 2011). Thanks to "hybrid models", i.e. the fusion of classical econometric methodologies and machine learning (Makridakis et al., 2018), it is possible to exploit such "big data" to obtain a forecasting performance superior to that of standard models (Li et al., 2017). By proposing an innovative modelling framework, we develop a hybrid model for the Lombardy tourism market, to verify the reliability of the ambitious target of becoming one of the most visited regions in Europe. We find mixed results for the Lombardy provinces, calling for a more refined and micro-founded tourism planning strategy. Furthermore, our framework is easily extendable with other countries and/or research areas, thus providing a valuable tool for assessing such issues.