Evaluation of machine learning models based on household food insecurity data in Indonesia

Household food insecurity is a critical issue, and accurate prediction models are essential for identifying at-risk households and guiding policy decisions to address this issue. This study compared the effectiveness and stability of two machine learning models: random forests (RF) and generalized random forests (GRF). Predicting household food insecurity using food insecurity experience scale data in West Java, Indonesia. The evaluation showed that the GRF model performed best and exhibited more consistent predictions. The important variables that influence household food insecurity in West Java are household size, type of house floor, bank savings account ownership, type of house wall, sanitation facility adequacy status, cash transfer program status, land ownership status, and food assistance recipient status.