Nieves et a. use the random forest machine learning method to predict what value globally? Describe in detail how random forest works. What is a dasymmetric population allocation? Which geospatial covariates proved to be the most important when predicting global values of where humans reside?
Random forest machine learning uses a tree method in which the program randomly creates a multitude of probability trees for all the possible options. The random forest machine learning method differs from other, more primitive, machine learning networks, in that it takes the mode of the results from the test in order to determine which outcome is the most effective and probable. Dasymetric mapping on the other hand, is a tool to determine population density relative to land use. It does this by using boundaries that divide the area into zones of relative homogeneity. The tool allows the user to see the population density broken down over the relative areas alongside other areal data. In regards to the covariates considered in this paper, covariates within urban/suburban areas were the most important, as they can often indicate density on their own without analysis. Additionally, transportation as well as public resources available to people were several other important covariates. Nieves et al. also included riverways and healthcare as two crucial covariates to account for as they both are signals of high population. This is because hospitals and other healthcare systems are designed to help as many people as possible, while rivers serve as water sources and transport source for those living near them.