Simultaneous Prediction of Multiple Dimensions of Food Security

Program officers are faced with difficult modelling choices when monitoring the state of diverse yet related dimensions of household food security. We compare results of single- and multi-output frameworks, applying both econometric and machine learning models, to jointly predict three interrelated outcomes of food security: prevalence of undernourishment (access), dietary diversity (utilization), and food market dependence (stability). Using the 2013–2022 Kyrgyz Integrated Household Survey, the results show that accounting for interdependencies among outcomes significantly improves accuracy, with the Generalized Structural Equation Model (GSEM) outperforming both single-output regressions and single- and multi-output neural networks. The asymmetry between caloric adequacy, dietary diversity, and market exposure explains that multi-output modeling enhances predictive power over single outputs. The findings can guide program officers navigating the trade-offs between prediction performance, interpretability and feasibility, showing that linear models can deliver robust predictions even when outcome dependencies are jointly addressed. These findings advance understanding on the interrelations between multiple dimensions of food security and have important implications for designing monitoring strategies that recognize the multidimensional nature.