Assessment of humanitarian crises and disaster risk exposure using data-driven Bayesian Networks

Abstract

Humanitarian crises and disasters are low-probability, high-impact risk events that can significantly jeopardize the sustainability of economic growth and development at a national level. The assessment of disaster risk can help policy-makers to ascertain the exposure and vulnerability of communities and enhance their capacity level to cope with such rare events. In this paper, we adopt a data-driven approach to assess the country-level risk associated with the humanitarian crises and disasters in a network setting. The contribution of this paper is to prioritize multi-dimensional factors contributing to the disaster risk exposure while capturing the network-wide impact of individual factors. Further, the factors are assessed for both enhancing and reducing the overall network-wide risk exposure. Governance and currentconflict intensity are considered as the most critical factors for the high-risk and low-risk countries, respectively, whereas projected conflict intensity is assessed as the most critical factor relative to its network-wide impact.