A 100-m gridded population dataset of China’s seventh census using ensemble learning and geospatial big data

A 100-m gridded population dataset of China’s seventh census in 2020 was generated by stacking ensemble learning and geospatial big data. The county-level and town-level data of China’s seventh census and ten related covariates at the 100-m resolution were first collected as the input datasets. Three popular machine learning algorithms (i.e., random forest, XGBoost, and LightGBM) were chosen as base models to create and train the stacking ensemble learning to generate gridded population dataset for China.The estimated gridded population dataset (R²=0.8936) is more accurate than existing WorldPop (R²=0.7427) and LandScan (R²=0.7165) products assessed by the town-level test census data. The dataset is associated with the paper of "Yuehong Chen, Congcong Xu, Yong Ge, Xiaoxiang Zhang and Ya'nan Zhou. A 100-m gridded population dataset of China’s seventh census using ensemble learning and big geospatial data, 2024" published in the journal of Earth System Science Data.

本数据集为基于堆叠集成学习（stacking ensemble learning）与地理空间大数据生成的中国2020年第七次全国人口普查100米格网人口数据集。
首先收集中国第七次全国人口普查的县级、镇级统计数据，以及10种分辨率为100米的相关协变量作为输入数据集。
选取随机森林（random forest）、XGBoost、LightGBM三种主流机器学习算法作为基模型，构建并训练堆叠集成学习模型，以生成中国区域格网人口数据集。
经镇级人口普查检验数据评估，本数据集的决定系数R²=0.8936，优于现有WorldPop（R²=0.7427）与LandScan（R²=0.7165）两款数据集产品。
本数据集关联发表于《地球系统科学数据（Earth System Science Data）》期刊的论文："Yuehong Chen, Congcong Xu, Yong Ge, Xiaoxiang Zhang and Ya'nan Zhou. A 100-m gridded population dataset of China’s seventh census using ensemble learning and big geospatial data, 2024"。