Optimizing Wheat Price Forecasting and Reporting through Data Pipelines, Machine Learning, and Automation

Abstract

This paper investigates the critical importance of wheat procurement strategies in countries heavily reliant on imports, such as Tunisia, where wheat imports are exclusively managed by the government entity, the Office of Cereal. Given the significant impact of price fluctuations on national budgets due to large import volumes, optimizing procurement strategies is essential. This study highlights the pivotal role of Free On Board (FOB) and freight prices in determining the final Cost and Freight (CNF) prices. The research begins with the collection and processing of FOB and freight price data, which is stored in a cloud-based dataset. A data pipeline is then developed to ensure continuous access to updated information. Machine learning models, including Random Forest and Support Vector Machines (SVM), are employed to forecast FOB prices. These forecasts, combined with freight data, are integrated into an interactive Power BI dashboard for data visualization and actionable insights. To further streamline access to updates, an automated reporting system using Power Automate is implemented, enabling the distribution of the latest information via email. This study demonstrates the effectiveness of combining machine learning, interactive visualizations, and automation to enhance procurement strategies and provide stakeholders with timely, data-driven insights for navigating the complexities of the global wheat market.