Machine learning price index forecasts of flat steel products

Investors and authorities have always placed a high emphasis on commodity price forecasting. In this study, the issue of daily price index forecasting for flat steel products on the Chinese market between June 15, 2011, and April 15, 2021 is examined. There hasn't been enough focus in the literature on anticipating this important commodity price indicator. Cross-validation and Bayesian optimizations are used to train our models, and Gaussian process regressions are used to support our conclusions. With an out-of-sample relative root mean square error of 0.1293%, the created models correctly forecasted the price indices between April 26, 2019, and April 15, 2021. The developed models may be used for policy research and decision-making by investors and policymakers. When creating similar commodity price indices with reference data on the price trends predicted by the models, the forecasting findings may prove helpful.