In addition, the ROI information can be employed to build possibly important, added, attributes for random forest predictive classifiers. These attributes can be employed to create a machine design that classify behaviors that rely upon the spatial location of system pieces in relation to the ROIs.