Cluster Analysis: Theoretical Development and Application to Socioeconomic Data of Ecuador

Abstract

This study proposes a socioeconomic vulnerability typology for Ecuadorian parishes based on official data from the 2022 Population and Housing Census. A total of 308 parishes are analyzed using 17 socioeconomic and basic services indicators, previously standardized. Hierarchical cluster analysis is applied using Ward’s method with Manhattan distance, evaluating multiple linkage criteria and distance metrics. The optimal number of clusters is determined through the agglomerative coefficient, silhouette width, and dendrogram inspection, resulting in a robust five-cluster solution. The validity of the classification is further assessed through a comparison with the K-means algorithm, achieving a 93% agreement between both clustering approaches. The resulting clusters reveal distinct territorial patterns in socioeconomic conditions and access to basic infrastructure, enabling the identification of priority areas for targeted public policy interventions. These findings demonstrate the usefulness of cluster analysis as a data-driven tool for territorial planning and decision-making, providing a reproducible and adaptable methodological framework applicable to other geographic and socioeconomic contexts.