Regional steel price index predictions for the southwest Chinese market through machine learning

Commodity price forecasts have long been a major source of reliance for both investors and governments. This study analyses the difficult task of projecting regional steel price indices recorded on a daily basis for the southwest Chinese market, using data spanning from 1 January 2010 to 15 April 2021. The forecast of this major commodity price indicator has received insufficient attention in the literature. Forecasts here are facilitated through Gaussian process regressions, which are estimated based upon cross-validation together with Bayesian optimisations. With an arrived-at relative root mean square error of 0.4508%, the constructed models accurately forecast the price indices out-of-sample from 8 January 2019 to 15 April 2021. Investors and government authorities can utilise established models to research prices and make relevant decisions. When reference data on the price trends proposed by these models are employed, forecasting findings might aid in the development of similar commodity price indices.