Predicting Sulphur Price Volatility: A Multi-Model Approach for Enhanced Commodity Forecasting

The volatility in sulphur prices presents significant challenges for industries relying on sulphur as a key raw material, particularly in terms of budgeting, inventory management, and strategic planning. Traditional methods for predicting commodity prices, such as manual analysis, are often time-consuming and error-prone. This project proposes a comprehensive predictive system for sulphur price forecasting by leveraging advanced machine learning and time series techniques. Multiple predictive models, including Linear Regression, Random Forest, AdaBoost, XGBoost, ARIMA, Auto ARIMA, and VAR, are implemented to develop accurate and reliable price forecasts. The project integrates diverse datasets such as crude oil prices, sulphur import data, and internal purchase records to enhance the robustness of the predictions. A rigorous evaluation of model performance using metrics like Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) is conducted to identify the best-performing model for sulphur price prediction. The outcome of the project is expected to deliver precise forecasts, enabling businesses to make informed procurement decisions and optimize their operational strategies. Additionally, the research contributes to the broader understanding of commodity price prediction methodologies by comparing machine learning and time series models in the context of sulphur pricing. This predictive system offers a significant improvement in the financial planning and risk management processes of industries dependent on sulphur, providing actionable insights and enhancing overall business performance.