Spectrum regression testing

This is a shiny app and package for comparing different runs of AIM via extract.

Installation

You can install the development version of {spectrum.regression.testing} like so:

remotes::install_github("AvenirHealth-org/spectrum.regression.testing")

Setup

This app uses data from 2 different sources

1. Manual extracts from dropbox
2. spectrum-orderly

This repo uses dotenv to manage configuration via a .env file. To run the ShinyApp we need to provide a path to locally synced or downloaded extract files, and a path to the spectrum-orderly root.

1. Sync the extracts folder from Dropbox Avenir Shared Drive/Projects/Models/AIM/Shiny regression testing note the local path.

2. Sync the orderly remote from Programming Sharepoint Documents/spectrum-orderly note the local path

3. run the ./scripts/initialize_data.R to create the .env file for you or write the paths into a file at the root of this repo .env e.g.

   EXTRACTS_DIR=C:\Users\Test\Avenir Health Dropbox\Avenir Shared Drive\Projects\Models\AIM\Shiny regression testing\extracts
   ORDERLY_ROOT=C:\Users\Test\Avenir Health\Programming - spectrum-orderly
   

Run

You can launch the application by running:

spectrum.regression.testing::run_app()

To run in development run the script ./scripts/run_dev.R or source the app in an R session with devtools::load_all() and then run the app with run_app()

Updating data

New orderly run

The orderly model fit task is run automatically once a week. You can manually run it by going to the spectrum-orderly-runner run AIM GitHub actions

1. Click the “Run workflow” at the top right below the list of workflow runs
2. Always use the “main” branch here, but you can select the version of spectrum desktop to install. This must match one of the release tags from Spec5 repo e.g. v6.42 or v6.43-beta.3. If not specified it will use the release tagged as “latest”
3. The CI will install the specified version of Spectrum and then run Extract via the CLI and save the results. The orderly task will also record the hash of the Spec5 repo for this Spectrum version in metadata. It records the Spectrum version in parameters. These can be used for querying downstream.

Note that running via the CI like this will only work for versions of Spectrum for which a release is available. If you want to run it before a release has been made then it is best to run the orderly task manually. See the spectrum-orderly repo for details of how to do this.

Updating the extract configuration

The extract configuration is stored in the orderly repo shared directory update this here the configuration is used in this ShinyApp to know what indicators are available for display. The configuration from the orderly run will be read and used automatically.

Displaying new indicators

Update the metadata/indicator-table.csv