In the supply chain industry, manufacturers’ ability to forecast prices of raw material is paramount to optimize production. A tailored buying strategy can reduce costs of production and increase revenue for manufacturers. Over the years, scientists have produced many techniques and models to address these challenges, giving birth to the science of Supply Chain and Industrial Engineering.

A recent paper published by a team of students in Industrial Engineering at UC Berkeley aims at forecasting a metal price listed in the commodity market for the next 10 weeks based on the historical prices. Using Mean Absolute Percentage Error (MAPE) to evaluate the results, the team reviewed famous models and designed new ones as part of a benchmark. Among those, the Linear Regression (MAPE 1.9%) and the traditional times series forecasting method ARIMA (MAPE 2.4%) fared better than XGBoost (MAPE 3.1%) and the promising deep learning GRU encoder-decoder (MAPE 7.1%).