The random forest machine learning method is used to predict human population density values in low and middle income countries. Random forest works by creating a “forest” of unrelated decision trees, which together, can make a prediction. Individual trees are “weak learners”, and are combined to make a “strong learner”. Dasymetric allocation involved allocating population distributions based on geospatial information, like land cover type. It reassigns population distribution that has just been assigned to boundaries and locations. The two most important covariates that proved to be important when predicting global values of where humans reside, from the article, are variables related to built/urban areas, and also climatic/environmental covariates. Other important variables were urban.suburban extents, populated place covariates, and transportation networks. To rank the importance of the variable, a measure of WIR was assigned - weighted importance rank. WIR is calculated by within-country ranked importance, divided by total number of covariates in the country model.