strandbygaard/rtemplateproject

Template R Project Structure

Do's and Dont's

Be stylish! Keep a consistent style to the code. Follow the tidyverse R style guide https://style.tidyverse.org/.

Initialize the project with a GIT repository Always initiate your project with at GIT repository, and use version control throughout.

And write sensible GIT commit messages to your future self Read this https://chris.beams.io/posts/git-commit/ :-)

Keep a tidy folder structure Use the following folder structure as a starting point

R Contains R source files. Why is it called R? Because devtools and R packages require this, so it's the most compatible naming convention.

config Contains all configuration data

data Contains all data used in the project

data/raw Contains raw input files as received

data/temp Contains temporary data that needs to be stored on disk

data/processed Contains processed raw data

tests Contains unit test cases for all functions

output Contains output from code

Use a build script Always use a build script to drive your code

Put functions in separate files Functions don't have to live in the same file as the calculations. Put them in a separate file, and include them using source()

But is it stylish? Use lintr to check your code for convention and obvious problems. Run it prior to every commit, e.g. lintr::lint("R/sample.R"). To show output from lintr in the Markers pane, select Code > Show Diagnostics from the menu. Add it as a dependent step in your build script.

Running tests

Use the following structure as a starting point:

1. Functions are defined in files under src
2. Create a test file in test for each file with functions that should be tested in

E.g.

1. Functions are defined in src/sample.R
2. A test file is created test/test-sample.R
3. Load testthat package with the command library(testthat) in the R console
4. Run the tests from the project root with the command test_file("./tests/testthat/test-sample.R") in the R console