An integrated economic and machine learning approach using DNN and SVR for forecasting Iran’s copper market under climate change

Climate change mitigation and carbon reduction form an imperative policy agenda for societies and governments in the 21st century. Demand for the key metals such as lithium, graphene, cobalt, and above all, copper—a basic element in renewable energy systems and sustainable infrastructure—has generated interest in undertaking deeper drilling operations to access untapped ore reserves, trends in production, and market patterns. A novel integration of Deep Neural Networks (DNN), Support Vector Regression (SVR), and Time-Varying Parameter Vector Auto regressions (TVP-VAR) has been employed to analyze and forecast copper price dynamics in Iran, in light of internal shocks and global decarbonization policies. Copper, as a critical input in renewable energy systems, is influenced by both geopolitical developments and environmental transitions. Major disruptions such as JCPOA-related sanctions, public health crises, and domestic unrest have been considered in modeling market responses. Sliding-window feature engineering and hyperparameter optimization have been applied to enhance model accuracy. The DNN model demonstrated strong sensitivity to stress scenarios (R² = 0.993, RMSE = 2,901), while SVR offered smoother long-term projections (R² = 0.865). These findings suggest that copper prices in Iran are influenced not only by economic factors but also by climate-related and political conditions. Policy recommendations are provided for aligning mining operations with low-carbon strategies. These include investment in copper futures, CCS technologies, and green infrastructure development to support resilience against sanction-driven volatility and climate risks. The framework serves as a data-driven tool for policymakers to stabilize the copper market in Iran and navigate a dual challenge of international isolation and environmental obligations.