Replication package for "Comparison of machine learning algorithms and model evaluation approaches for deforestation risk mapping in the Brazilian Amazon"

Replication package for "Comparison of machine learning algorithms and model evaluation approaches for deforestation risk mapping in the Brazilian Amazon." Risk_maps.zip contains the risk maps created based on the different models tested. data.csv contains the data (sampled pixels) used for model fitting. def_risk.R is the R code used for model fitting (some editing is needed to run the code in a different computer/directory). mapping_risk.R is the R code used to create the risk maps (some editing is needed to run the code in a different computer/directory). risk_on_prodes_pixels.R is the R code for the validation method based on all deforested pixels (some editing is needed to run the code in a different computer/directory). map_stack.zip contains the underlying data/maps used for model fitting and risk prediction. Additional analyses were performed in Dinamica EGO, available at https://csr.ufmg.br/dinamica/