Big Data analytics for forecasting tourism destination arrivals with the applied Vector Autoregression model

The prediction of tourist numbers is important for Destination Management and Marketing. While most existing methods rely on well-structured statistical data, using web search queries of the destination to forecast its tourist arrivals is a new way to apply Big Data analytics. However, there are no studies exploring correlation of weather, temperatures, weekends and public holidays with tourism destination arrivals and web search queries of the destination, respectively. This study uses the Vector Autoregressive modeling to examine the Granger causality between actual arrivals of the studied cultural tourism destination and its web search queries, and to explore the correlation mentioned above. The striking result is that weather has no correlation either with actual arrivals of the studied cultural tourism destination, or with its web search queries. Meanwhile, unlike previous researchers who discuss the predictive power of web queries on actual tourism flows, this study emphasizes their reciprocal predictive powers upon each other. The originality of this study is exemplifying the utilization of Big Data analytics in the tourism domain with Big Data datasets, data capture techniques, analytical tools, and analysis results. This study further digs possible reasons for an identified short time lag length (p = 2), to provide insights for Destination Management and Marketing. •Finds that weather and temperatures have no correlation with actual arrivals of the destination•Finds that weekends and public holidays have significant correlation with the actual arrivals•Emphasizes the tourism destination arrivals' predictive power on its web search queries•Provides an example of Big Data analytics for Destination Management and Marketing