We propose a supervised learning approach to statistically quantify the magnitude of extreme events on vulnerable communities using publicly available panel data directly reflective of the different dimensions and manifestations of social suffering. The manifestations along these dimensions...