Random Forest Population Mapping Complexity Reduction Algorithm, Data and Code

Though the randomForest R package provides the functionality to fit a model with an arbitrarily large number of covariates and observations (limited only by memory and disk space) a limiting feature of our approach is the time spent during the prediction phase.

This code and sample data provides the details of a data reduction method that greatly increases the prediction-phase for new data, necessitated by running per-pixel predictions on large countries for WorldPop population mapping products.