Predicting wholesale edible oil prices through Gaussian process regressions tuned with Bayesian optimization and cross-validation

Purpose - Developing price forecasts for various agricultural commodities has long been a significant undertaking for a variety of agricultural market players. The weekly wholesale price of edible oil in the Chinese market over a ten-year period, from January 1, 2010 to January 3, 2020, is the forecasting issue we explore. Design/methodology/approach - Using Bayesian optimisations and cross-validation, we study Gaussian process (GP) regressions for our forecasting needs. Findings - The produced models delivered precise price predictions for the one-year period between January 4, 2019 and January 3, 2020, with an out-of-sample relative root mean square error of 5.0812%, a root mean square error (RMSEA) of 4.7324 and a mean absolute error (MAE) of 2.9382. Originality/value - The projection's output may be utilised as stand-alone technical predictions or in combination with other projections for policy research that involves making assessment.