Figure 2: Marginal effect curves of each environmental predictor in the ensemble BRT model fitted to the full data set.
From: Predicting the risk of avian influenza A H7N9 infection in live-poultry markets across Asia

The shaded areas represent the density of the predicted relationships to each environmental correlate (with the effect of the other correlates marginalized) from all 120 sub-models, within the lower and upper 95% quantiles of the distribution. The solid lines give the mean effect curves calculated from all models. Sub-plots are ordered by the mean of their RC to each sub-model, with these average RCs given in parentheses with each sub-plot.