Country-survey level stunting, wasting, and underweight burden, 1983–2019, with the same model-derived drivers reported in the accompanying paper. Built as an interactive prototype of the kind of monitoring dashboard a Programme Monitoring and Improvement team would maintain in Tableau, Power BI, or Salesforce.
About this dashboard: data current as of the 2019 survey round in the underlying panel (compiled 2021). Interactive in the browser (filters re-render every chart and figure immediately); not connected to a live data feed. Uses public secondary data only and is not produced, reviewed, or endorsed by GAIN.
Averaged across country-survey observations in the current filter selection. Dashed line marks the WHO high-burden threshold used in the paper (20% for stunting, 15% for wasting).
Mean |SHAP| from the trained XGBoost stunting-burden classifier reported in the paper. This ranking is a fixed property of the trained model and is not recomputed by the filters above, so the dashboard and the paper always agree on these numbers.
USAID Data Quality Assessment dimensions applied to the panel feeding this dashboard. Static across filters, since these are properties of the source data, not of any one slice of it.