Smart Growth Strategies A Machine Learning Framework for Population-Driven Water and Energy Demand Forecasting in Urban Planning

This project presents a machine learning framework for forecasting per capita energy consumption using population-driven urbanization and climate metrics to support smart urban planning and sustainable resource management. The framework integrates Random Forest and Decision Tree models to analyze global urbanization patterns and predict energy demand (energy_kg_oil_eq_cap) based on key indicators such as total population, population density, GDP, urban and rural population percentages, access to electricity, safe sanitation access, renewable energy consumption, and CO₂ emissions. The dataset global_urbanization_climate_metrics.csv, covering the period 1960–2023 across multiple countries, underwent comprehensive preprocessing including mean imputation for missing values, removal of non-predictive columns, and exploratory data analysis with a correlation heatmap to uncover important relationships. Feature selection was performed using Mutual Information Regression, retaining the nine most influential variables for model training. The system employs supervised regression models, with the Random Forest Regressor delivering superior performance (Mean Squared Error: 391,052.34, R² Score: 0.896) compared to the Decision Tree Regressor (Mean Squared Error: 782,753.62, R² Score: 0.792). Both models were trained on a 67-33 train-test split and saved for deployment. By effectively capturing complex non-linear relationships between urbanization dynamics and energy consumption, the framework enables accurate demand forecasting that can guide strategic decisions in infrastructure development and resource allocation. This approach offers urban planners and policymakers a powerful, data-driven tool for anticipating future energy needs driven by population growth, supporting sustainable smart city initiatives, optimizing urban-rural resource distribution, and evaluating the long-term impact of development policies. The project highlights the potential of machine learning in transforming urbanization and climate data into actionable insights for resilient and sustainable urban growth.

Keywords: Urban Planning, Energy Demand Forecasting, Population Growth, Smart Cities, Random Forest, Decision Tree, Sustainable Development, Urbanization, Predictive Analytics, Feature Selection.