UCLA

Parametric Hawkes models are proposed and fit by maximum likelihood to World Health Organization data from the 2014 Ebola epidemic in West Africa. Models were fit to various sub-region-level subsets of the data to compare with previous research on compartmental models and nonparametrically estimated Hawkes processes. Models were also fit to country-level subsets and multi-country subsets to evaluate how these models perform on increasing scales. Results suggest that these spatio-temporal models are able to accurately forecast the spread of Ebola infections on larger space-time windows than have been previously researched, with the benefit of improved parameter interpretability.