Population spatialization in Zhengzhou city based on multi-source data and random forest model

Fine-scale population map plays an essential role in numerous fields, including resource allocation, urban planning, disaster prevention and response. Point of Interest (POI) data is widely used for spatialization, but the types POI are ignored. Since different have impacts on distribution, this paper typed other multi-source to distributions at fine scales. At township level, three random forest models were generate maps 150 m, 300 500 m 2020, enabling downscaling county-level distribution grid level. The main influencing factors extracted analyzed based feature importance output from model. Zhengzhou city was as a case study. experiments show results spatialization all scales study better fitting accuracy than that GPWv4 LandScan datasets. coefficient determination ( R2 ) 0.8333 gridded population, 0.8295 0.8224 m; related residence information greater contributions features; more conducive spatialization.