Hybrid V-Support Vector Regression as Statistical Methodology for Forecasting Daily Natural Gas Prices

The simultaneous accurate forecast of natural gas prices together with London Gas Oil prices leads to optimal market functioning. The complex nature of predicting natural gas prices and London Gas Oil prices becomes very challenging because of their non-linear and non-stationary historical price patterns. v-support vector regression(v-SVR) functions as an established method for predicting metal price variations. The v-SVR achieves better performance results through adjusting its hyperparameters because these values determine its operational effectiveness. v-SVR requires optimal hyperparameter selection for achieving high accuracy levels in its predictions. The research suggests using beluga whale optimization (BWO) as a meta-heuristic framework to enhance the v-SVR hyperparameter estimation which can predict the prices of natural gas and London Gas Oil. The forecasting performance evaluation used multiple metrics including DM test combined with R2, DA, RMSE,MSE, IA and MAE. The prediction capabilities of BWO-v- SVR surpass those of the competitor methods GS-SVR and CV-v-SVR. . Multiple evaluation criteria demonstrated that the proposed BWO-v-SVR technique shows positive outcomes when compared to different available options.