Prediction of US Agricultural Commodity Prices with Ordinary Least Squares and Long Short-Term Memory Neural Networks

Abstract

Commodity price prediction for agricultural commodities is a crucial research topic in the field of agriculture for realizing the sustainable and healthy development of agriculture. The agricultural commodities prices are influenced by irregular fluctuations may be due to global warming; so that, it may affect global food security. This research aims to provide crucial information for the development of agricultural commodities price prediction. The weekly price data of US agricultural commodities: Wheat, Sugar, Cocoa, Coffee, and Cotton from the last 10 years is mainly investigated. This data prediction using the Ordinary Least Squares and the LSTM neural networks is compared in terms of the mean square error and the mean absolute error. As a result,  it can be clearly seen that the performance of LSTM neural networks outperformed in predictions of commodities prices according to OLS.