Using Graphical Models for Missing Data Patterns Detection in Sustainability Surveys

Ensuring sustainability in the agri-food sector requires com- prehensive data analysis. This study examines missing data patterns in a large-scale survey of Italian agri-food companies within the Agritech National Center, focusing on sustainability variables. The basic idea is that failure to provide a value for these variables indicates little attention to the issue of sustainability. We employ graphical models to infer the con- ditional independence structure among missingness indicators, which are modeled as binary variables. We consider two classes of graphical mod- els: graphical log-linear models and graphical Ising models. The graphical Ising models allow us to handle a larger set of variables, integrate fully observed variables. The models are applied to the Agritech data.