Mobility is an important metric in modeling of community population dynamics and community resilience. It is directly associated with the inorganic changes in a community during and after a disruption (e.g., city gentrification, refugee migration from a war zone, flash mobs in an online community, etc.). Mobility is driven by socioeconomic, demographic, geographical, psychological, and legal parameters. Not all of these parameters are mutually independent (orthogonal). For proper modeling, it is important to avoid collinearity, as otherwise the model will not generalize well. We discuss how machine learning can be used to avoid it by identifying the mutually orthogonal metrics (factors)