Multidimensional Determinants of Poverty and Regional Clustering in North Sumatra, Indonesia: A Factor and Cluster-Based Analytical Approach

Abstract

This study investigates the multidimensional nature of poverty and its spatial distribution across 33 districts and cities in North Sumatra, Indonesia. Using a combination of Principal Component Analysis (PCA) and K-Means clustering, the research identifies key socioeconomic factors contributing to regional poverty disparities, including education, unemployment, housing quality, and local fiscal capacity. The clustering results reveal five distinct district typologies, ranging from urban centers with strong infrastructure and human capital to rural and island regions facing structural deprivation. Multiple regression analysis confirms that education level, unemployment, and uninhabitable housing significantly predict poverty levels. The findings highlight the limitations of one-size-fits-all poverty alleviation strategies and underscore the need for geographically targeted policies. This study provides empirical evidence to support region-specific planning under Indonesia’s decentralized governance framework and offers a scalable approach for other provinces facing similar socio-economic diversity.