Title:
Multidimensional Analysis of Poverty Through Topological Data Analysis

Authors:
Oscar A. Jiménez, Felipe A. Llaugel, Dominican Republic

Abstract:

This study addresses the phenomenon of poverty in low income community in Dominican Republic, from a multidimensional perspective, moving beyond the limitations of traditional income-only approaches. The central objective was to characterize the empirical structure of social deprivation by applying unsupervised learning techniques and Topological Data Analysis (TDA). Data from 591 households were processed. Using the Mapper algorithm together with DBSCAN, two candidate metric spaces were evaluated and compared to capture the “shape” of the data: a prevalence-weighted semimetric and the standard Euclidean metric. Unlike linear aggregation methods, the topology built on the Euclidean metric proved better suited to this micro-territorial context: it revealed that poverty in this sector does not operate in isolated clusters but instead forms an uninterrupted structural continuum that closely tracks the Quality of Life Index (ICV). These findings show that the absence of certain items responds strongly to sociocultural priorities and local consumption patterns. 

DOI: http://dx.doi.org/10.51505/ijaemr.2026.11501