Towards an automated framework for agent-based simulation of refugee movements

Abstract

Forced migration is a growing global problem, and the world now has a record amount of 22.5 million refugees. Models that predict refugee movements are few and far between, and constructing these models requires a substantial amount of manual effort while erupting refugee crises require a very rapid response. Here we present a vision towards establishing an automated framework, aimed to enable researchers to construct simulations of refugee movements more quickly and systematically. Our approach incorporates a diverse range of data sources, and uses the FabSim toolkit in conjunction with the Flee simulation code to quickly generate simulation workflows. In addition, we highlight a few key steps that we have already taken towards realizing this vision and discuss opportunities for wider applicability.