Forecasting Coal Prices Using Decision Tree Approach

Abstract

Coal is an important fuel in the world's energy systems. Today's world depends heavily on coal. It contributes to 41% of the world's power and acts as a key ingredient in the production process of 90% of cement and 70% of the world's steel. An accurate prediction model for forecasting future coal prices would provide critical information and a timely caution to governments to ensure a steady energy supply. Studies on forecasting coal prices regarding related commodities such as crude oil, iron, natural gas, and paper are rare. Hence, this research aims to forecast coal prices using historical data for the 2013-2016 period. The study uses a decision tree model to anticipate the coal prices, and the impulse response function and variance decomposition are used to show the dynamic connections of coal prices with various variables. The model in predicting coal prices performs well with mean absolute percentage errors of around 5%. The study results will likely enable new insights in providing valuable inputs for all the stakeholders.Keywords: Coal price forecasting, decision tree, prediction