A Two-Stage Dimension Reduction and Clustering Framework for Financial Behavior and Socio-Demographic Profiling

Abstract

Financial inclusion (FI) is a critical component of global financial advancement. Despite technological progress, about 1.5 billion people in emerging economies lack access to formal financial systems. Understanding financial decision-making behaviors and preferences is essential for enhancing comprehension of financial inclusion. This study focuses on Peru, a country with relatively low FI levels. Using data from the 2019 National Survey of Demand for Financial Services and Financial Literacy (NSDFS), we implemented a two-stage clustering methodology with dimension reduction techniques and clustering algorithms to uncover social profiles within the Peruvian population. Our findings identified three clusters based on financial behaviors and socio-demographic characteristics. The optimal configuration, utilizing Isomap reduced to 2 dimensions combined with the K-means++ algorithm, achieved a mean aggregated score of 0.832, yielding the best results among the other dimension reduction and clustering techniques considered. The clusters highlighted disparities in financial access, emphasizing the need for targeted interventions. These insights can aid policymakers and regulators in developing strategies to enhance FI in Peru, underscoring the value of clustering techniques in addressing financial inclusion challenges.