While anticipating the outcomes of natural disasters is a growing area of interest to a diverse body of organizations, little formal work has been done to make these predictions for specific countries. We propose a novel methodology that combines supervised learning on historical disaster and demographic data with intelligent clustering to expand a relevant training set for particular countries and disaster types. By clustering countries at snapshots in time, based on developmental stages and ‘disaster event profiles,’ we develop more specific sets of historical data that improve predictions on the number of people killed and financial cost of various disaster types.
How to Cite:
Mandair D. & Griffiths L. & Liu C., (2017) “Predicting Societal and Economic Impact of Future Natural Disasters”, University of Arizona Journal of Medicine 1(2). p.21-25.