Interval Forecasting Using GEV Regression With Stochastic Search Variable Selection: A Simulation Study and Its Application to Forecast Imported Meat Price Range

This study aims to forecast the price range, i.e., the interval forecast between the upper and lower bounds of prices, using Generalized Extreme Value Regression (GEVReg) as a robust tool with significant implications for the trade industry. The price range is crucial for identifying undervalued or overvalued imported goods. However, predicting this price range is challenging due to the asymmetric distribution of import prices, non-stationary trends, and time-varying covariates. Numerous covariates, including lags of response variables, increase computational complexity and multicollinearity issues, requiring variable selection to identify the most impactful covariates. To address the challenges mentioned earlier, this work proposes GEVReg-SSVS, which combines the GEVReg as an underlying model with Stochastic Search Variable Selection (SSVS), a data-driven approach to variable selection. In addition, a simulation study was conducted to evaluate the performance of the GEVReg-SSVS method, which shows that it excels in variable selection, consistently identifying models with covariates aligned with the true generated model. These insights offer a robust approach for forecasting imported meat prices while detecting potential price anomalies, directly benefiting the trade industry.