Data for Development

Data analysis, automation and evidence-based decision-making with AI-augmented R workflows

What you will build

Over five days you move from a raw PDHS .DTA file to a published, reproducible report. Not five disconnected topics — one pipeline, assembled a piece at a time.

The analytical arc

1. Acquire & wrangle data that arrives messy — files, APIs, PDFs (Day 1)
2. Measure & see — indicators, weights, CPI splicing, tables and charts (Day 2)
3. Model & map it, with the right weights and the right CRS (Day 3)
4. Report it in Quarto — parameterised, so it re-runs next month (Day 4)
5. Present it, after one hour of revision (Day 5)

AI assistance runs through every day, not a day of its own — see the prompt library.

What you need

• R ≥ 4.3 and Positron (or RStudio)
• Quarto ≥ 1.4
• The workshop data pack (data/), including HIES 2024-25, PDHS, WFP prices and PBS CPI
• No prior R required — Day 1 starts from zero

See Setup before Day 1.

Start here

Agenda	Day-by-day session plan and timings
Setup	Install R, Positron, Quarto and the packages
Day 1	Foundations, data acquisition and the AI toolkit
Prompt library	AI prompts, used every day
Reference	Cheat sheets to keep open while you work

How the material is organised

Each day folder holds three kinds of file:

• index.qmd — the session plan for that day
• NN-*.qmd — slides and walkthroughs, in teaching order
• scripts/*.R — the live-coding scripts, runnable on their own

Every script resolves data with here::here("data/..."), so you can run any script from anywhere in the project without changing your working directory.

TipWorking alongside the facilitator

Open the project by double-clicking d4d.Rproj. That sets the project root, which is what here::here() anchors to. Opening a loose .R file instead will break paths.