<p>Forecasts regarding the prices of agricultural commodities have long been significant to a range of market stakeholders. Our study looks at weekly wholesale price indices of green grams in an attempt to solve the challenge. This price index has important economic implications and the analyzed sample spans a 10-year period from 01/01/2010 to 01/03/2020. Here, price index forecasts are produced by applying Gaussian process regression methods, which are derived using Bayesian optimization techniques and cross-validation procedures. Our empirical prediction technique yields quite accurate price index projections for the out-of-sample period spanning from January 5, 2018, to January 3, 2020, as suggested by a relative root mean square error of 2.6273%, a root mean square error of 4.3178, and a mean absolute error of 2.9583. These promising forecast results imply that the Gaussian process regression, as a machine learning technique, might have good potential for other research into price forecasts of different commodities. Price prediction models, such as Gaussian process regressions constructed here, provide governments and investors with the knowledge they need to make informed decisions, including but not limited to setting future pricing strategies, determining terms of trading contracts, developing market risk management plans, and designing policies.</p>