Analyzing Multidimensional Poverty in Roma Settlements: A WEFE Nexus and Machine Learning Approach

Abstract

TThis study aims to analyze multidimensional poverty determinants within Roma
settlements in Serbia, North Macedonia, and Montenegro using logistic
regression and machine learning to identify the socioeconomic and resourcebased components influencing poverty within the WEFE (Water Energy Food
Ecosystems) Nexus. Reliable lighting, sanitation maintenance, safe water access,
consistent water supply, and energy for cooking all play a critical role in poverty
alleviation. Our key findings align with SDG 1 (No Poverty), SDG 6 (Clean
Water and Sanitation), and SDG 7 (Affordable and Clean Energy), underscoring
the global significance and relevance of our research. Random Forest and Extra
Trees perform very well when compared to logistic regression by capturing
highly variable interactions that may be missed by logistic regression. Results
with country-specific emphases are presented, such as digital access in
Montenegro and household size in North Macedonia, to illustrate the adaptability
of the WEFE framework to different regional contexts. The results advocate for
resource-driven integrated policies to improve people’s access to important
utilities, financial inclusion, and digital connectivity to build social sustainability
and resilience. The study supports NexusNet’s plan to lead SDG-aligned poverty
reduction in all sectors in the Western Balkans by focusing on WEFE resources
and socio-economic supports.