Geographical clustering of the electoral flow between the Italian general elections of 2018 and 2022

This paper aims to group Italian regions based on their similarity in voting flows between the major political coalitions during the 2018 and 2022 General Elections. We employ a Fuzzy C-Medoids clustering approach, using a tailored dissimilarity measure that captures the differences between matrices while also addressing the presence of outliers to mitigate their impact. The objective function integrates spatial constraints via fuzzy modularity, enabling the model to consider spatial relationships among regions. The methodology is applied to the ITANES dataset, a panel sample of 4696 respondents revealing interesting patterns, most notably a clear dichotomy between northern and southern Italy.