Docker Desktop for Statisticians: Running R in Containers

Image by Author | ChatGPT

Docker Desktop transforms how statisticians work with R by providing isolated, reproducible environments that eliminate version conflicts and setup headaches. Instead of installing R directly on your computer and managing package dependencies, you can run R inside containers that come pre-configured and ready for analysis. If you are new to Docker and would like some introduction on common terminology and how to get started, please review our article on “Understanding Docker for Statisticians: Containers, Images, and Reproducible Analysis.”

Navigating Docker Desktop’s Interface

With Docker Desktop running, you’ll see a clean interface with several key sections in the left sidebar. The Images section shows your downloaded software templates, while Containers displays your running environments. Think of Images as software installers and Containers as the actual running programs.

The main view provides helpful learning resources and tutorials for new users. At the bottom, you’ll notice resource usage indicators showing how much memory and CPU your containers are consuming. This real-time feedback helps you understand Docker’s impact on your system.

Pulling the Official R Image

Let’s get R running in a container. Click on Images in the sidebar, then click the blue “Search images to run” button. In the search box, type r-base and press Enter.

You’ll see the official R image appear in the results, marked with the “OFFICIAL IMAGE” badge. This image contains a complete R installation maintained by the R Foundation. Click the Pull button next to r-base:latest.

Docker Desktop will download the image, showing progress bars as it retrieves the R software and all its dependencies. This might take a few minutes depending on your internet connection. Once complete, you’ll see r-base appear in your local images list with a green status indicator.

Starting Your First R Container

With the R image downloaded, click on it in your Images list. You’ll see detailed information about the image, including its size and available tags. To start the container, click on the Terminal tab on the bottom right corner and provide the below one liner:

# Start the container
docker run -it --rm r-base:latest R --vanilla

Your new container appears in the Containers section with a green “Running” status. Click on the container name to open its details view. Here you’ll find several tabs including Logs, Inspect, and Files. Click the Terminal tab to access R directly.

Running Statistical Operations in R

In the terminal window, you’ll see the familiar R prompt (>). You’re now running R inside an isolated Docker container, completely separate from any R installation on your host computer.

Let’s perform some basic statistical operations:

# Create a sample dataset
data <- c(23, 25, 28, 30, 32, 35, 38, 40, 42, 45)

# Calculate descriptive statistics
mean(data)
sd(data)
median(data)

These commands work exactly as they would in any R installation, but you’re running them in a containerized environment. Try a more complex statistical test:

# Generate two sample groups
group1 <- c(12, 15, 18, 20, 22, 25, 28)
group2 <- c(16, 19, 22, 24, 27, 30, 33)

# Perform a t-test
t.test(group1, group2)

The t-test results appear just as they would in RStudio or any other R environment. The key difference is that this R session is completely isolated from your main computer.

Understanding Container Isolation

To demonstrate isolation, let’s start a second R container. Return to your Images list and start a fresh terminal. Configure it the same way and start the container.

Now you have two R sessions running simultaneously, each in its own container. In the first container, create a variable:

my_analysis <- "Container 1 data"

Switch to your second container’s terminal and try to access this variable:

print(my_analysis)

You’ll get an error because the variable doesn’t exist in the second container. Each container maintains its own isolated R environment, preventing conflicts between different analyses or experiments.

Container Management Through the GUI

Docker Desktop makes container management intuitive. In the Containers section, you can:

Stop containers by clicking the stop button (square icon)
Start stopped containers with the play button
Delete containers you no longer need
View logs to see all output from your R sessions
Access files created within the container

Try stopping one of your R containers and starting it again. Notice that any variables you created in the R session are lost – containers start fresh each time. This behavior ensures reproducible environments but means you’ll want to save important work to files.

Why This Matters for Statistical Work

Container isolation solves common statistical computing problems. You can test experimental R packages without affecting your main installation, run legacy analyses with older R versions, or share your exact computational environment with collaborators.

Each container starts with a clean R installation, ensuring reproducible results. When you share your analysis code, you can also share the Docker image, guaranteeing that others get the same R version and computational environment.

This approach transforms statistical computing from “hope it works on your machine” to “guaranteed identical environments everywhere.”