Validation for Split-Population Duration IED Models
Abstract
How generalizable is split-population duration modeling of violence at small scales? This proposal is a follow-up to an earlier project that developed a predictive model of IEDs using split-population duration regression and SIGACT data for two areas in Afghanistan in 2009 and early 2010. To validate the models developed in this effort, we propose to examine two additional grids in Afghanistan, as well as areas in Colombia and the Philippines. In addition to assessing how stable the performance of split-population duration regression is across these new contexts, it should show what factors need to considered when adjusting resolution limits with new data, and how to adjust for varying background intensities of conflict across difference areas of the world.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141512348
Entities
People
- Michael E Ward
Organizations
- Duke University
- Office of Naval Research
- United States Navy