Forecasting copper prices by decision tree learning

Abstract

Forecasting the prices of metals is important in many aspects of economics. Metal prices are also vital variables in financial models for revenue evaluation, which forms the basis of an effective payment regime using by resource policymakers. In this article, we utilized a machine-learning algorithm based on decision tree to predict future copper prices. We showed that our method is capable of accurately and reliably predicting copper prices in both short-term (days) and long-term (years), with mean absolute percentage errors below 4%. We also demonstrated that the current method is assumption-free, robust, and not prone to human bias. This method is easily and readily applicable to predicting the prices of other metals and other commodities, and we expect that such method could be useful in a broad range of fields.