Cleans and deidentifies Liberian community-health referral spreadsheets so PII is removed before analysis.
# liberia-deidentify
Cleans and deidentifies Liberian community-health referral spreadsheets so
PII is removed before analysis. Standardizes worker names against a master
list, hashes every name column to a stable opaque id, and drops the
direct-PII columns (patient names, phone numbers, baby birth date). The
cleaned CSV is safe to share; the salt and the id↔name reverse-lookup
table stay local.
## Setup (first time only)
You need **R** (the language) and **RStudio** (the editor). Both are free.
Install in this order:
1. **R** — go to
cloud.r-project.org, click the download link
for your operating system (Windows or macOS), and run the installer
with the default options. R must be installed *before* RStudio.
2. **RStudio Desktop** — go to
posit.co and download the free
*RStudio Desktop* installer for your operating system. Run it with
the default options.
Then download this project folder (`liberia-deidentify`) to your computer
and keep all the files together in one folder.
To run a script: open RStudio, use *File → Open File…* to open one of the
`.R` files from the project folder, and click the **Source** button in
the top-right of the editor pane (or press `Ctrl+Shift+S` on Windows,
`Cmd+Shift+S` on macOS).
**The first time you Source any script**, R will automatically download
and install the packages it needs. You'll see lines like
`Installing required package: openssl` scrolling in the Console pane
below the editor. This takes 1–3 minutes and needs an internet
connection. Subsequent runs skip the install and start immediately. No
separate OpenSSL or other system-level install is required on Windows or
macOS — everything is handled by R's package installer.
If R asks you to choose a CRAN mirror, pick any one near you (the scripts
also set a default mirror, so this prompt usually doesn't appear).
## Workflow
1. **Build the master list** (`extract_unique_names.R`) once per data drop.
Reads worker rosters and wri …