francescochecchi/ken_malnut_prediction

R scripts and source data to replicate predictive modelling of acute malnutrition burden in drought-prone regions of Kenya ## Predicting the burden of acute malnutrition in drought-prone regions of Kenya: a statistical analysis ### Description of input data and R analysis scripts September 2024 ## Input data The following input datasets are included in the `/in` directory of the repository: * `surveys.zip`, which users should unzip such that all files therein are contained within a `/in/surveys` sub-directory. Each unzipped .csv file contains a single survey dataset, though some datasets contain multiple explicit survey strata. FIle names should not be altered. All datasets are fully anonymised; * `ken_predictor_data.xlsx`, each worksheet of which contains a single predictor dataset. the `metadata` tab lists each worksheet (predictor) and its characteristics, and is read by R so as to then read the individual worksheets. More detail on each dataset is provided in the paper; * `ken_other_data.xlsx`, which contains in each worksheet other datasets needed for the analysis, again listed in a `metadata` worksheet; * `admin_boundaries.zip`, which users should unzip to a `/in/admin_boundaries` sub-directory. This file contains shape files delineating county and sub-county boundaries for Kenya; A few additional predictor datasets are not uploaded to the repository due to their large size. Users should therefore download them first, as follows: * normalised difference vegetation index (NDVI) data should be downloaded from gmes.icpac.net to a `/in/ndvi` sub-directory. The data platform requires users to sign up and request specific data for download, which they will then receive an email link for. Users should select the 'Monthly NDVI' and 'Monthly SNDVI (Anomaly)' products from 2014 to 2019 in GeoTiff format for all of Kenya. The download will be a single zipped file. Users should unzip this file to the above sub-directory. * Data on estimated access to safe assisted births should be downloaded from hub.worldpop.org to a `/in/mnh_acc …