CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: AMH copula-based clustering of variables with application to district heating demand Authors:  F Marta L Di Lascio - Free University of Bozen-Bolzano (Italy)
Andrea Menapace - Free University of Bozen-Bolzano (Italy)
Roberta Pappada - University of Trieste (Italy) [presenting]
Abstract: Understanding thermal consumption in urban areas is a crucial need to increase the sustainability and efficiency of energy systems and reduce the impact of climate change. The focus is on district heating, which represents one of the key technologies involved in the ongoing process aimed at reducing the waste of energy in a flexible urban energy system. Motivated by the features of high-frequency district heating demand data in the Italian city of Bozen-Bolzano, we develop a clustering methodology grounded on a copula-based dissimilarity measure in the hierarchical framework. To this aim, we exploit the Ali-Mikhail-Haq copula to cluster residential users (buildings) based on the observed time series of heating consumption. The copula approach allows us to tackle both temporal and cross-sectional dependence while considering the effect of meteorological variables and spatial information. We investigate the proposed dissimilarity measure through Monte Carlo studies and compare it with its analogue based on Kendall's rank correlation. In particular, we evaluate the performance of the two measures in terms of overall dendrogram quality and agreement between the two partitions obtained. The application of the proposed measure to district heating demand yields clusters of buildings that are homogeneous with respect to their main characteristics, such as energy efficiency and heating surface, thus providing crucial information for the study of sustainable energy scenarios.