Predicting Civil Unrest: Venturing Through Latin America

• Activity Number: 33
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Andrew Hoegh*+ and Scotland Leman
• Companies: Virginia Tech and Virginia Tech
• Keywords: event modeling ; Dynamic Linear Models ; spatio-temporal modeling ; Poisson Processes ; CAR models ; Gaussian Processes

Abstract

Civil unrest is an interesting phenomena to model. Certain conditions need be in place (e.g. government corruption) in order for a society to be ripe for civil unrest; however, these conditions alone do not warrant protests. In some cases, such as the "Occupy Movement" or the "Arab Spring Revolutions" protests spread rapidly across a country. In other cases, like a factory strike, protests are relatively localized and do not give rise to other protest. The factors leading to the spread of civil unrest include spatial proximity at the country, state, and city level, the type of protest, and the population protesting. In this work we develop a spatio-temporal model for civil unrest in Latin America. Specifically, dynamic linear models are introduced that incorporate correlation in the "who", "what", and the "where" components of the protests in order to map baseline levels and capture trending protests.