Skip to main content
. 2018 Feb;24(2):284–293. doi: 10.3201/eid2402.170704

Figure 3.

Figure 3

Partial dependence plots for the 4 most influential variables in boosted regression tree (BRT) model for antibodies against Vi capsular antigen of Salmonella enterica serovar Typhi, Fiji, 2013. A) Age; B) distance to major rivers and creeks; C) annual rainfall; and D) distance to flood-risk areas. The final ensemble BRT was constructed with 50 BRT models and 11 environmental and social covariates by using data from 1,305 samples. Gray areas indicate 95% CIs of plots.