Technical Bulletin No. (TB-1959) 25 pp

November 2022

Evaluating the Prediction Performance of the International Food Security Assessment’s Production Models: A Cross-Validation Approach

The International Food Security Assessment (IFSA) model of the U.S. Department of Agriculture’s (USDA) Economic Research Service (ERS) was developed to help USDA and its stakeholders evaluate the food security outlook of low- and middle-income countries for up to 10 years. The IFSA model provides an estimate of total food demand and food production, both elements in measuring food security. This research evaluates the production side of the IFSA model to determine the best performing prediction model specification. Accurate estimations and projections of the domestic production that accounts for geographic and crop level differences will help increase the accuracy of the projections and ultimately enhance the supply and demand dynamics in the IFSA assessment.

Keywords: cross-validation, global food security, production, prediction, mean squared error, root mean squared error, mean absolute error

In this publication...