Interactive R development

APPLIES TO: Azure CLI ml extension v2 (current) Python SDK azure-ai-ml v2 (current)

This article shows how to use R in Azure Machine Learning studio on a compute instance that runs an R kernel in a Jupyter notebook.

The popular RStudio IDE also works. You can install RStudio or Posit Workbench in a custom container on a compute instance. However, this has limitations in reading and writing to your Azure Machine Learning workspace.

Prerequisites

• If you don't have an Azure subscription, create a Trial before you begin. Try the free or paid version of Azure Machine Learning today
• An Azure Machine Learning workspace and a compute instance
• A basic understand of using Jupyter notebooks in Azure Machine Learning studio.

Run R in a notebook in studio

You'll use a notebook in your Azure Machine Learning workspace, on a compute instance.

1. Sign in to Azure Machine Learning studio

2. Open your workspace if it isn't already open

3. On the left navigation, select Notebooks

4. Create a new notebook, named RunR.ipynb

   Tip

   If you're not sure how to create and work with notebooks in studio, review Run Jupyter notebooks in your workspace

5. Select the notebook.

6. On the notebook toolbar, make sure your compute instance is running. If not, start it now.

7. On the notebook toolbar, switch the kernel to R.

   

Your notebook is now ready to run R commands.

Access data

You can upload files to your workspace file storage resource, and then access those files in R. However, for files stored in Azure data assets or data from datastores, you must install some packages.

This section describes how to use Python and the reticulate package to load your data assets and datastores into R, from an interactive session. You use the azureml-fsspec Python package and the reticulate R package to read tabular data as Pandas DataFrames. This section also includes an example of reading data assets and datastores into an R data.frame.

To install these packages:

1. Create a new file on the compute instance, called setup.sh.
2. Copy this code into the file:

#!/bin/bash

set -e

# Installs azureml-fsspec in default conda environment 
# Does not need to run as sudo

eval "$(conda shell.bash hook)"
conda activate azureml_py310_sdkv2
pip install azureml-fsspec
conda deactivate

# Checks that version 1.26 of reticulate is installed (needs to be done as sudo)

sudo -u azureuser -i <<'EOF'
R -e "if (packageVersion('reticulate') >= 1.26) message('Version OK') else install.packages('reticulate')"
EOF

1. Select Save and run script in terminal to run the script

The install script handles these steps:

• pip installs azureml-fsspec in the default conda environment for the compute instance
• Installs the R reticulate package if necessary (version must be 1.26 or greater)

Read tabular data from registered data assets or datastores

For data stored in a data asset created in Azure Machine Learning, use these steps to read that tabular file into a Pandas DataFrame or an R data.frame:

1. Ensure you have the correct version of reticulate. For a version less than 1.26, try to use a newer compute instance.

   packageVersion("reticulate")
   

2. Load reticulate and set the conda environment where azureml-fsspec was installed

   library(reticulate)
   use_condaenv("azureml_py310_sdkv2")
   print("Environment is set")
   

3. Find the URI path to the data file.

   1. First, get a handle to your workspace

      py_code <- "from azure.identity import DefaultAzureCredential
      from azure.ai.ml import MLClient
      credential = DefaultAzureCredential()
      ml_client = MLClient.from_config(credential=credential)"
      
      py_run_string(py_code)
      print("ml_client is configured")
      

   2. Use this code to retrieve the asset. Make sure to replace <MY_NAME> and <MY_VERSION> with the name and number of your data asset.

      Tip

      In studio, select Data in the left navigation to find the name and version number of your data asset.

      # Replace <MY_NAME> and <MY_VERSION> with your values
      py_code <- "my_name = '<MY_NAME>'
      my_version = '<MY_VERSION>'
      data_asset = ml_client.data.get(name=my_name, version=my_version)
      data_uri = data_asset.path"
      

   3. To retrieve the URI, run the code.

      py_run_string(py_code)
      print(paste("URI path is", py$data_uri))
      

4. Use Pandas read functions to read the file or files into the R environment.

pd <- import("pandas")
cc <- pd$read_csv(py$data_uri)
head(cc)

You can also use a Datastore URI to access different files on a registered Datastore, and read these resources into an R data.frame.

1. In this format, create a Datastore URI, using your own values:

   subscription <- '<subscription_id>'
   resource_group <- '<resource_group>'
   workspace <- '<workspace>'
   datastore_name <- '<datastore>'
   path_on_datastore <- '<path>'
   
   uri <- paste0("azureml://subscriptions/", subscription, "/resourcegroups/", resource_group, "/workspaces/", workspace, "/datastores/", datastore_name, "/paths/", path_on_datastore)
   

   Tip

   Instead of remembering the datastore URI format, you can copy-and-paste the datastore URI from the Studio UI, if you know the datastore where your file is located:

   1. Navigate to the file/folder you want to read into R
   2. Select the elipsis (...) next to it.
   3. Select from the menu Copy URI.
   4. Select the Datastore URI to copy into your notebook/script. Note that you must create a variable for <path> in the code.

2. Create a filestore object using the previously mentioned URI:

fs <- azureml.fsspec$AzureMachineLearningFileSystem(uri, sep = "")

3. Read into an R data.frame:

df <- with(fs$open("<path>)", "r") %as% f, {
 x <- as.character(f$read(), encoding = "utf-8")
 read.csv(textConnection(x), header = TRUE, sep = ",", stringsAsFactors = FALSE)
})
print(df)

Install R packages

A compute instance has many preinstalled R packages.

To install other packages, you must explicitly state the location and dependencies.

For example, to install the tsibble package:

install.packages("tsibble", 
                 dependencies = TRUE,
                 lib = "/home/azureuser")

Load R libraries

Add /home/azureuser to the R library path.

.libPaths("/home/azureuser")

Once the libPath is updated, load libraries as usual.

library('tsibble')

Use R in the notebook

Beyond the issues described earlier, use R as you would in any other environment, including your local workstation. In your notebook or script, you can read and write to the path where the notebook/script is stored.

Next steps

• Adapt your R script to run in production