Data, rasters and code for supporting the Risk Map of Dengue Transmission in the Metropolitan Area of Mexico City.
# install.packages("remotes")
# remotes::install_github("fdzul/dengueriskcdmx")
install.packages("dengueriskcdmx", repos = c('https://fdzul.r-universe.dev', 'https://cloud.r-project.org'))
| Object | Description |
|---|---|
ae_aegypti_2017_2024 |
Ae. aegypti ovitrap presences and pseudo-absences, 2017-2024, with the split column (train / test) |
ae_aegypti_2017_2024_train |
Training partition (80%) |
ae_aegypti_2017_2024_test |
Test partition (20%) |
ae_aegypti_2025 |
External validation data, 2025 |
aoi |
Study area (Mexico City Metropolitan Area) |
Each data set has the class (presence / pseudoabs), nine predictors (built, tree, ndvi, prcp, ia, ice, im, pop, hfp) and the point geometry. The partition is 80/20, stratified by class, with seed 12345.
The Quarto code of the machine-learning workflow (ml_code_ae_aegypti.qmd and ml_code_dengue.qmd) is provided as vignettes.
Three rasters (about 500 m) are shipped in inst/extdata, all derived from the consensus (mean of the probabilities) of Random Forest, LightGBM and XGBoost models:
layer |
Raster |
|---|---|
"aedes" |
Probability of Ae. aegypti occurrence |
"dengue" |
Probability of confirmed imported dengue cases |
"risk" |
Five-level epidemiological risk |
library(dengueriskcdmx)
# ovitrap data
table(ae_aegypti_2017_2024$class)
# rasters (requires the terra package)
r <- ae_raster("aedes")
terra::plot(r)
The ovitrap data are included. The data on confirmed dengue cases used to fit the case model are not published; only the derived probability and risk rasters are provided.
An interactive version of the maps is available at https://eclectic-alfajores-088839.netlify.app/.
Dzul-Manzanilla F, Correa-Morales F. Risk Map of Dengue Transmission in the Metropolitan Area of Mexico City (under review).
MIT