Capstone: Reproducible Spatial Pipeline
The capstone takes a district-level health indicator from raw data to a clustered, mapped, and interpreted result, as a reproducible pipeline rather than a one-off script.
What you build
A pipeline on district-level health data (the bundled AHS/NFHS example, or an indicator of your choice):
- Profile the raw data:
capstone/R/02-profile-document.R - Clean & validate:
capstone/R/03-clean-validate.R - Tidy & reshape:
capstone/R/04-tidy-reshape.R - Transform (derive indicators):
capstone/R/05-transform.R - EDA & plot:
capstone/R/06-eda-plot.R - Analyse (model and spatial clustering):
capstone/R/07-analyse.R
Run the chain with capstone/run-all.R. It produces the figures and tables that capstone/report.qmd knits into a report.
For the clustering step on its own (Queen contiguity, spatial weights, Moran’s I, LISA), see capstone/ahs_clustering_exercise.r.
Deliverable
Pick a public health indicator (stunting, wasting, underweight, anaemia, IMR) and submit:
- the reproducible script(s) or Quarto document,
- a global and local clustering analysis (Moran’s I and a LISA cluster map),
- a 500-word interpretation written for a public health decision-maker.
The capstone scripts write their own outputs/ (data, figures) when you run capstone/run-all.R. Run it once before knitting capstone/report.qmd.