Spatial dynamics across food systems transformation in IFAD investments: a machine learning approach

The food systems approach has gained renewed prominence in recent years, due to its role towards gaining an understanding of food insecurity and malnutrition. A “food systems” lens has therefore become essential to better design development interventions and innovations that can positively impact food systems outcomes. This study provides evidence on the dynamics across food system dimensions within development projects supported by the International Fund for Agricultural Development (IFAD). A custom taxonomy was developed and machine learning techniques primarily focused on supervised text mining, network analysis and LASSO regression were applied to IFAD project documentation to extract analytics about food systems’ spatial and temporal thematic representation over 40 years of project implementation. The paper thus provides insights about the dynamics as well as transformations of food systems within IFAD’s stated activities, providing a historical overview of how the Fund has tackled food systems over four decades of project life cycles. Findings show an overall increase in reporting against food system dimensions and consolidate the applicability of machine learning analytics to uncover trends about international agencies’ activities and accelerate knowledge generation around strategic themes.