Supplementary Data for: A Multi-Sensor Machine Learning Framework for Union-Level Spatial Estimation of Childhood Stunting in Bangladesh: Addressing Spatial Disaggregation in Data-Constrained Settings

This dataset contains the processed analytical data and administrative boundary shapefiles used in the study titled "A Multi-Sensor Machine Learning Framework for Union-Level Spatial Estimation of Childhood Stunting in Bangladesh: Addressing Spatial Disaggregation in Data-Constrained Settings," published in Spatial and Spatio-temporal Epidemiology. The dataset includes: 1. FINAL_master_cluster_data_manuscript.csv - Processed analytical dataset derived from the 2022 Bangladesh Demographic and Health Survey (BDHS) and satellite-derived environmental variables. - Contains 670 DHS cluster locations with stunting prevalence estimates and five environmental predictors: NDVI, land surface temperature (LST), nighttime lights, rainfall, and elevation. - Variables: cluster_id, stunting_rate, child_count, mean_haz, severe_stunting_rate, ndvi_mean, lst_mean, nightlights_mean, rainfall_mean, elevation_mean, longitude, latitude, district. 2. bgd_admin4 shapefiles - Administrative boundary shapefiles for Bangladesh at the union level (admin level 4). - Includes 5,160 union polygons covering the entire country. - Files: bgd_admin4.shp, bgd_admin4.shx, bgd_admin4.dbf, bgd_admin4.prj, bgd_admin4.cpg. - Coordinate Reference System: EPSG:4326 (WGS84). These data support the reproduction of union-level spatial predictions of childhood stunting prevalence across Bangladesh and the generation of high-resolution maps presented in the manuscript. Data sources: - Demographic and Health Survey (DHS) Program: https://dhsprogram.com/ - MODIS NDVI and LST: NASA EarthData - Nighttime Lights: NOAA VIIRS - Rainfall: CHIRPS, Climate Hazards Center - Elevation: SRTM DEM, USGS - Administrative boundaries: Bangladesh Bureau of Statistics (BBS) and Humanitarian Data Exchange (HDX)