Food security is a critical global issue, especially in low-income countries where volatile food prices endanger access to nutritious food. Despite global efforts, gaps in food price forecasting leave vulnerable populations exposed to sudden price shocks, impacting health and livelihoods. While existing tools offer valuable data, they often lack accessible, interpretable forecasts, limiting their effectiveness in supporting equitable food security solutions. This project aims to develop the Agricultural Commodity Analysis and Forecasts (AGRICAF) methodology to help address these gaps.   

AGRICAF is distinguished by its use of publicly available data, ensuring that it remains accessible and relevant to a broad audience, including governments, agricultural cooperatives, and non-governmental organizations. The methodology comprises four main stages:

  1. Data Collection and Screening: This stage involves gathering time series data on AC prices and other explanatory variables from global open-sources like the World Bank, FAO, and USDA. The data undergoes rigorous testing for stationarity and suitability for analysis, ensuring consistency and reliability.
  2. Retrospective Analysis and Variable Selection: In this stage, AGRICAF performs a retrospective analysis using various XML models. The analysis identifies key variables driving price shifts, using regression for annual relative changes and extensive cross-validation techniques to ensure robustness. The methodology reduces the number of impactful variables, refining the dataset for accurate forecasting.
  3. Price Forecasting: The refined dataset is used to forecast AC prices for up to 12 months using XML and time series models. AGRICAF employs rolling cross-validation to enhance forecast accuracy, capturing complex relationships in the data without relying heavily on prior assumptions.
  4. Interpretation of Forecasts: The final stage focuses on making the forecast results accessible to non-specialist users. AGRICAF employs various model-agnostic techniques, to explain the impact of different variables on price changes.
Artwork created using DALL-E © Artwork created using DALL-E

AGRICAF offers a robust, accessible, and interpretable solution for forecasting global agricultural commodity prices. Its ability to deliver accurate forecasts using publicly available data makes it an essential tool for supporting food security strategies, particularly in low-income countries. The methodology's potential to democratize access to critical market information can help reduce the knowledge gap in global agricultural trade, contributing to a more equitable and sustainable food system.

The project is funded by the European Commission in the framework of the Marie Skłodowska-Curie Actions (MSCA). These Postdoctoral Fellowships offer researchers holding a PhD the opportunity to gain new skills and experience while carrying out their own research project abroad. Rotem Zelingher will complete this fellowship via placement in Austria over a term of 24 months. 

AGRICAF marks a major step towards achieving the Sustainable Development Goal of ending hunger by equipping vulnerable communities with the tools to navigate global food markets more fairly. By democratizing food price forecasts, AGRICAF promotes food justice, social equity, and well-being, ensuring critical market insights are accessible to all.