Ali-Mikhail-Haq copula to detect low correlations in hierarchical clustering

In this work we introduce a new dissimilarity measure based on the AliMikhail-Haq copula, motivated by the empirical issue of detecting low correlations and discriminating variables with very similar rank correlation. This issue arises from the analysis of panel data concerning the district heating demand of the Italian city Bozen-Bolzano. In the hierarchical clustering framework, we empirically investigate the features of the proposed measure and compare it with a classical dissimilarity measure based on Kendall’s rank correlation.