Forecasting Big Data Daily Silver Price Based on Hybridizing Support Vector Regression and Meta-Heuristic Algorithms

Because commodities are prone to large price swings, forecasting their prices can be extremely difficult. Similar to other commodities, the price of silver is determined by a complicated interplay between a number of factors, including time-lag. Precise forecasting of the silver price is essential for the stable and effective functioning of silver markets. The time series that metal prices follow is non-stationary, nonlinear, and contains periods that fluctuate due to possible growth. Accurate forecasting of silver prices is difficult since they are consistently highly nonlinear and non-stationary. Forecasting metal prices, in general, have been predicted by support vector regression (SVR). The SVR is a modification of the recently developed machine learning and statistical theory-based classification paradigm. However, the SVR performance is highly sensitive to the choice of its hyperparameters that usually needed to tune. Therefore, choosing such hyperparameters is an essential component of SVR. This paper proposing employing the golden jackal optimization algorithm (GJO) algorithm, a meta-heuristic approach, to improve SVR hyperparameters determination and as a result improving silver prices’ forecasting. Depending on the several forecasting criteria, our findings demonstrate that the suggested approach outperforms two benchmark approaches and can significantly raise the silver price prediction’s accuracy. These findings have significance for investigating artificial intelligence in the commodities market and for quickening the recovery of the global economy.