cd2030.pooled

The Countdown to 2030 Pooled app: a Shiny app that combines several countries' saved datasets (the .rds files the RMNCAH and Vaxx apps save) into one pooled file, then lets you explore it, compare countries and export tables.

It runs inside DataSuite (through the countdown-analytics extension) or on its own in R. How this package fits with the others, and how a release reaches users: countdown-analytics/docs/ARCHITECTURE.md.

Install and run

install.packages("cd2030.pooled", repos = c("https://damurka.r-universe.dev", "https://cloud.r-project.org"))
cd2030.pooled::run_app()                                   # prints (runs) the app
cd2030.pooled::run_app(pre_loaded_dir = "saved-datasets/")

run_app() returns a Shiny app object; printing it or shiny::runApp() runs it. Its arguments default to the environment variables DataSuite sets when it launches the app:

Argument	Environment variable
`pre_loaded_dir`	`CDSUITE_SHINY_SELECTED_FILE`	a folder of saved datasets (`.rds`) to start from (DataSuite asks for a folder); `NA` for none
`language`	`CDSUITE_SHINY_LOCALE`	`"en"`, `"fr"` or `"pt"`
`app_name`	`CDSUITE_SHINY_NAME`	the name in the header
`app_version`	`CDSUITE_SHINY_VERSION`	the version in the header; defaults to this package's version

In DataSuite nothing needs installing by hand: DataSuite installs this package when the extension is installed and updates it when the extension is updated.

What it does

Page
Build pooled file	1. choose the data domain (RMNCAH or Immunization), what to include (everything, or the standard tables only) and the files or a folder; 2. load each (loaded or failed, with the reason); 3. review; 4. create one pooled `.rds` holding every dataset.
Explore pooled data	open a pooled file (the one just built, or another); a page per kind of dataset (parameters, overall score, indicator coverage, coverage, mortality, service utilization, ...) with its data and graphs; Compare countries and All datasets across the whole file; Extract a piece: chosen datasets, countries, years and columns. Exports as CSV, Excel or `.rds`.

Unlike RMNCAH and Vaxx it doesn't use cd2030.core's cd_app() (there is no single dataset to load): it builds its own two pages on the same shell (cd_app_ui(), cd_shell_server()), in the green "pooled" theme. It reads each country's file with cd2030.core.

File	What
`R/run_app.R`	`run_app()`: translations, the nav, the Build pages, the server
`R/pooled-datasets.R`	the tables a pooled file can hold and how each is pulled out of one country's dataset (plain R)
`R/pooled-build.R`	loading the files and combining them
`R/pooled-kinds.R`	the kinds of dataset Explore has a page for
`R/pooled-explore.R`, `R/pooled-compare*.R`, `R/pooled-charts.R`	the Explore pages, comparing countries, the charts
`R/pooled-export.R`	CSV, Excel and zip exports
`R/pooled-ui.R`	small UI pieces of this app
`inst/translation/translation.json`	the app's own texts
`inst/www/pooled.css`	the app's styles (served at `cd-pooled/`)
`app.R`	runs the app from this folder (see below); not part of the package

Develop

shiny::runApp()          # from this folder: app.R loads the source with pkgload::load_all()
devtools::check()        # must stay at 0 errors, 0 warnings, 0 notes

app.R uses pkgload::load_all() when it finds DESCRIPTION next to it, otherwise the installed package. Keep R/ ASCII (\uXXXX escapes in strings); column names used unquoted in dplyr code go in R/globals.R.

When cd2030.core changes what a dataset holds, check R/pooled-datasets.R: it is where each table is pulled out.

Release

Bump Version: in DESCRIPTION, add a NEWS.md entry, devtools::check() clean.
Commit, tag vX.Y.Z, push main and the tag. r-universe builds it within about an hour (instantly with its GitHub app installed).
DataSuite users get it at the next update of the countdown-analytics extension. If they must have this version, raise package.version for Pooled in the extension's package.json and release the extension.