Cluster Validation in Unsupervised Machine Learning with Application to the Analysis of the Tourism Demand in Italy after COVID-19 Lockdown

Abstract

COVID-19 has destroyed the tourism market, and it was essential to analyze the impact of the pandemics to reconstruct the tourism sector. In this work, we analyze the tourism demand in Italy using Social Big Data. In particular, we investigate the search behavior related to relevant tourism issues to identify some relevant partitions of Italy's demand. The analysis was conducted using unsupervised learning techniques in which the validation of the techniques was very relevant to identify the influential groups. The relevant results from the validated clusters obtained indicate no direct association between income and tourism choices. In this case, the worries about COVID-19 could be considered relevant to tourism choices. However, we observe a possibility for short-run growth for some clusters identified.

Keywords: COVID-19, Coronavirus, Machine Learning, Unsupervised Learning, Clustering Validation, Clustering, Tourism, Social Big Data, Tourism Demand

JEL Classification: C38, Z3, Z39