Understanding Food Insecurity Experience in Papua: An Explainable Artificial Intelligence Approach

This study examines household food insecurity in the Papua Island region of Indonesia using an explainable artificial intelligence (XAI) framework. The data used in this study are 20,669 households’ data from the March 2024 National Socio-Economic Survey (Survei Sosial Ekonomi Nasional/SUSENAS). Food insecurity is measured with Food Insecurity Experience Scale (FIES) and classified into two categories: food secure and food insecure. Separate classification models are developed for each of the six provinces using a random forest (RF) model combined with SHapley Additive exPlanations (SHAP) and followed by hierarchical clustering to examine how different factors shape households’ food insecurity status. The results of SHAP values cluster reveal clear spatial differences in the structure of vulnerability. In Papua, West Papua, and Southwest Papua, food insecurity is more associated with market dependency, financial constraints, and access to social assistance. In contrast, in Central Papua, Highland Papua, and South Papua, food insecurity is closely linked to dietary and nutritional intake and reliance on local agricultural production. These findings underline that food insecurity in Papua Island is not only a matter of food availability but also of unequal access, infrastructure gaps, and differentiated livelihood systems. The results highlight the need for geographically tailored policies while maintaining education improvement in Papua.