Year: 2021
Type: report
Abstract: Granular spatial information on the distributions of human population is relevant to a variety of fields like health, economics, and other areas of public sector planning.This paper applies ensemble methods and aims at assessing their applicability to analyzing and forecasting population density on a grid level.In a first step, we use a Random Forest approach to estimate population density in the Philippines and Thailand on a 100 meter by 100-meter level.Second, we use different specifications of Random Forest and Bayesian model averaging techniques to create forecasts of the grid-level population density in three Thailand provinces and evaluate their predictive power.
DOI: 10.22617/wps210519-2
Institution: Asian Development Bank
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4200112981