There seems to be a subconscious deterrent from using specialised tools like R as part of a production or core tech stack, as if it’s somehow not compatible with other forms of technology. (View Highlight)
Which is clealy BS. Over the past year a suprising number of my projects have involved ways to package and integrate complex R code into an existing application. (View Highlight)
Docker addresses these challenges by containerizing applications along with their dependencies, ensuring that they run consistently across environments. For R, this means you can create a portable, self-contained environment that includes your scripts, libraries, and configurations—all packaged in a lightweight container. (View Highlight)
Set Up Your R Environment
Before diving into Docker, start by setting up your R project. Ensure that:
• Your scripts are modular and well-organized. Below we have a basic setup where the script or code you want to run is in the app.R file.
• Dependencies are explicitly listed. I suggesting using {renv}, its good practice in general but it also makes installing the required packages a lot easier later on when we write our Dockerfile. (View Highlight)
Example or a bare-minimum directory structure:
my-r-project/
│
├── R/ # other R scripts or function
├── data/ # Data files (if applicable)
├── Dockerfile # Docker configuration file
├── renv.lock # Dependency file
├── app.R # Entry point
└── README.md # Documentation (View Highlight)
Key Notes on This Dockerfile:
• Base Image: We use the rocker/r-ver image, which is optimized for R and maintained here.
• System Dependencies: Add system-level dependencies your R packages might require.
• Install R packages: Notice how we are replying on renv’s automated package restoration? (View Highlight)
If you are doing serious work, you will likely want to move past cute ASCII art. In these cases you should invest a little more time into ensuring your R workflow is fit for purpose. Here are some resources to help go to the next level:
• If you want an easy guide to writing better, production quality R code check out my e-book.
• In particular, if you want to deploy a predictive model endpoint I cover this in the following case study.
• For a simple example of deploying R as a shiny app using a base Docker image check this repo out.
• You can even set up your own development environment using Docker. Just follow these steps. (View Highlight)
