Analyzing Electoral Data Using Partitional and Hierarchical Clustering Algorithms

Abstract

This paper describes the use of two clustering algorithms for investigating how results from the second round of the last Brazilian presidential election in Brazil are organized, taking into account values of the Municipal Human Development Index (MHDI). The investigation intended to uncover possible relationships between indicators that characterize profiles of Brazilian voters and profiles of Brazilian municipalities. MHDI is a customized version of the Human Development Index (HDI), and represents the development and quality of life offered by Brazilian municipalities. The work carried out is based on data results of the last presidential elections in Brazil, held in 2018, focusing on municipality´s data described by MHDI sub-indexes, as well as municipality’s population, related to the 5,558 municipalities in the country. The analysis of results of the second turn of the last presidential election, taking into account MHDI values was carried out based on clustering results induced by two clustering algorithms that employ different strategies: hierarchical (algorithm DIANA) and partitional (algorithm k-Means).