Enhancing Copper Price Prediction: A Machine Learning and Explainable AI Approach

This research introduces a hybrid model for copper price prediction, employs advanced machine learning models (linear regression, random forest, SVM, Adaboost, ARIMA), and utilizes the SHAP method for model interpretability. The study focuses on transportation-related variables over a 10-year period from Bloomberg Terminal, employing STL decomposition for time series forecasting. Key features impacting copper prices are identified, emphasizing the significance of demand, transportation, and supply. The Random Forest model highlights the critical role of demand. Addressing transportation supply constraints is crucial for enhancing model output in the dynamic copper market.