Skip to main content
Fig. 4 | Infectious Diseases of Poverty

Fig. 4

From: Neglected tropical diseases risk correlates with poverty and early ecosystem destruction

Fig. 4

Comparison between models’ performance and each relative variable importance. Overall performance and relative variable importance between ecological niche models and environmental (E) and socioenvironmental (SE) model predictors. A Small circles correspond to mean pROC values, generated based on a 5% error threshold, derived by multiple models using different algorithms (Random Forest, Maxent, SVM, and GLM). The diamond shapes represent the values from the ensemble (averaged) models for each disease. Blue represents models made with environmental sets of predictors and orange colors represents models made with socioenvironmental sets predictors. The dashed line corresponds to neutral performance. B The relative variable importance of each predictor is represented by circle size, in orange the models were made with environmental set predictors, in blue with socioenvironmental sets of predictors. The relative variable importance varies from 10 to 60%. pROC: Partial receiver operating characteristic; Br spotted fever: Brazilian spotted fever; temp.: temperature; precip.: Precipitation; Mean GDP PPP: Mean gross domestic product using purchasing power parity rates from [43]; % of households without toilet: Percentage of households without toilet from [38]

Back to article page