Scrap Steel Price Forecasting Using Time-Series Models with Exogenous Variables

This study proposes a quantitative forecasting framework for short-term scrap steel price prediction to support procurement and inventory planning in the steel industry. Monthly industrial data spanning a 12-year period are combined with economically relevant commodity, energy, and market indicators to capture the underlying drivers of scrap steel price dynamics. A systematic preprocessing procedure is applied, including missing-value screening and correlation-based feature selection, to reduce dimensionality and mitigate overfitting. Time-series regression models with exogenous variables are evaluated using standard forecast accuracy measures, namely Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination (R²). The empirical results indicate that the proposed ARIMAX-based model substantially outperforms a baseline price-only model, achieving notable reductions in forecasting error while improving explanatory power. Short-term forecasts over a 2–3 monthshorizon are generated together with 95% confidence intervals to explicitly quantify forecast uncertainty. While the model effectively captures medium-term price movements driven by energy costs and steel market conditions, short-term price volatility remains challenging due to abrupt market shocks and unobserved qualitative factors. Overall, the propose