New study reveals vulnerability of graphical modeling to data contamination.
Graphical modeling is a useful tool for understanding complex relationships between variables. Researchers studied how different methods can be affected by errors or unusual data points. They focused on models that include both discrete and continuous variables.