Comparison of Spatial Downscaling Methods for Population Density Estimation Based on Machine Learning
Population spatial distribution is key to urban planning and resource allocation,typical-ly understood through census data.The study aims to enhance the simulation accuracy of popula-tion census data's spatial distribution by comparing spatial downscaling methods.Utilizing multi-source data from remote sensing and socio-economic statistics,the study establishes spatial downscaling estimation models for Henan Province's 2020 census data using three machine learning techniques:Random Forest,XGBoost,and LightGBM.Results show LightGBM has the highest accuracy,followed by Random Forest and then XGBoost.The research not only supports spatial analysis of Henan's population but also offers insights for spatial downscaling of other statistical data.
NETL
NSTL
万方数据
