The tourist patterns in European countries. Application of a protocol for dynamic clustering of geographic spaces

Abstract

The methodological proposal presented in this paper enables a comprehensive multivariate exploratory analysis of time series taken in different geographical contexts. More specifically, it is a four-stage protocol for the identification of different dynamic patterns and the subsequent formation of geographic clusters. The clustering strategy compares the shapes of the trajectories using a natural, unsupervised approach based on learning from data. The exploratory techniques used in this method, which are readily applicable in any area of study, begin by reducing the dimensionality of a three-way data array. The potential of the strategy design is illustrated through a study of tourism dynamics in 27 European countries over the period 2007-2019. The profiles found for the trend and seasonal components reveal five distinct tourism patterns with some singularities. The choice of this sector for empirical research is not accidental. Tourism is connected to areas related to both economic and social activity. That is why their correct measurement and analysis are so interesting and necessary.