Machine learning scrap steel price forecasts for the regional east Chinese market

Purpose Investors and governments have traditionally depended on estimates of prices of various commodities. Using data from 08/23/2013 to 04/15/2021, this study aims to investigate the challenging problem of forecasting scrap steel prices that are published on a daily basis for the east China regional market. The research has not given much attention to predictions of this important commodity price indicator. Design/methodology/approach Gaussian process regression models, which are estimated using cross-validation approaches with Bayesian optimizations, are used to provide price forecasts. Findings Having a relative root mean square error of 0.4357%, the constructed models appropriately generate price forecasts for the out-of-sample testing stage from 09/17/2019 to 04/15/2021. Originality/value Models designed to research prices can be used by governments and investors to make well-informed decisions.