Assessing Behavior Similarity of Mineral Raw Material Prices through a Feature-Based Clustering Approach

Mineral raw materials prices have been shown to be affected by macroeconomic factors such as aggregate demand and commodity-specific factors (e.g., supply shocks). In addition, it has been shown that certain mineral raw material prices co-move, meaning that they behave similarly during expansion and contraction phases of the international business cycles. In order to assess the behavior similarity of the prices of different mineral raw materials, we propose a method that utilizes extracted features of time series price data and unsupervised learning techniques to create clusters of price movements having similar long-term behavior.