Identifying Patterns of Regional Polarization in Russia: A Machine Learning Approach

The study of regional polarization is critically relevant for Russia, as pronounced socio-economic disparities threaten national economic stability, social cohesion, and the strategic goal of technological sovereignty. These imbalances hinder efficient resource allocation and create vulnerabilities amid global realignment. This study analyzes patterns and drivers of regional polarization in Russia by leveraging machine learning to model the complex interplay of economic and non-economic factors. The research is guided by three hypotheses: polarization is driven by multi-dimensional disparities (H1); processes of convergence and divergence coexist (H2); and machine learning can effectively identify latent polarization structures overlooked by traditional methods (H3). K-means clustering identified regional typologies, with optimal cluster count validated by the Elbow Method and Silhouette Score. Predictive power and key drivers were analyzed using ensemble methods, Random Forest and XGBoost, with performance evaluated by Mean Squared Error (MSE). Results confirm a deeply polarized landscape with four distinct socio-economic clusters: resource-rich regions plagued by inequality and outmigration; leading hubs facing over-centralization risks; industrial-agrarian regions with persistent poverty; and lagging republics trapped in subsidy dependence. Analysis validated H2, showing simultaneous catch-up growth in some areas and entrenched divergence in others. Random Forest showed superior predictive accuracy (MSE = 0.132), confirming H3 and identifying key drivers (H1) beyond economic metrics. Theoretical significance lies in its novel, data-driven typology of Russian regions, highlighting the superiority of ML ensemble methods for modeling complex spatial inequality. Practically, findings provide policymakers with an evidence-based roadmap for differentiated regional strategies and a predictive tool for proactive intervention vital to balanced national development.