Machine Learning Approaches for Predicting Regional Growth and Urban Expansion

Abstract

Rapid urbanization requires accurate predictive models to support sustainable regional planning. This study proposes a Random Forest–based machine learning framework to predict regional growth and urban expansion using key indicators, including GDP growth, population density, infrastructure index, and night-time light intensity. The model demonstrates strong performance, achieving high accuracy (91%), R² of 0.94, and low RMSE (0.32), indicating robust predictive capability. Results show that infrastructure index, night-time light intensity, and population density are the most influential factors driving urban expansion, while GDP growth plays a secondary role. The model effectively captures non-linear relationships and produces predictions closely aligned with actual values across regions. The findings highlight the importance of spatial and infrastructural variables in shaping urban growth patterns. Methodologically, this study contributes a reproducible and interpretable framework, while practically offering insights for urban planning and policy formulation. The approach supports data-driven decision-making and promotes more efficient resource allocation. Future research should explore integration with remote sensing data, hybrid machine learning models, and spatio-temporal analysis to enhance predictive performance.