Skip to content

Posit Workbench Server (RStudio)

The Statistics Workbench server is web based development environment for R and Python. This server is intended for developing code only: any computationally intensive processes should be run on the Statistics Cluster.

Access

Workbench access is available to:

  • Statistics Computing Account users
  • Students enrolled in a Stat/CSSS course working with R. Access must be requested by the instructor two weeks prior to the start of the quarter.

To use Workbench, login to https://rstudio.stat.washington.edu using your NetID.

When you create a new session, you will prompted to select either RStudio for R, or JupyterLab for Python.

File Management

  • The Workbench home directory shares the Madrid home directory for anyone with a Statistics Computing Account.
  • For students using Workbench as part of a class, the Workbench home directory is independent of other Statistics home directories.

Files can be uploaded using the Workbench web interface for all users, and transferred by SFTP/SCP to madrid.stat.washington.edu:~ for Computing Account users (see Using SSH).

Cluster users can access their cluster home directory by navigating to /mnt/beegfs/homes/<netid>.

RStudio

R Versions

RStudio allows switching between R versions using the drop down menu in the top right. Currently these R versions available:

  • 4.3.2
  • 4.2.1
  • 4.1.2
  • 3.6.3

Please contact help@stat.washington.edu if you need another version of R installed.

Publishing to the Web with Workbench Connect (Shiny Server)

Rmarkdown and Shiny apps developed in RStudio can be published to the web using the Statistics Workbench Connect server. For more information, see Workbench Connect.

JupyterLab

Installing Packages

Python packages need to be installed with pip, then built into a custom JupyterLab environment called a kernel. In this example, the seaborn package will be installed:

  1. From a notebook, install the package with pip: !pip3 install seaborn
  2. Create Jupyter Python kernel. This will include any packages you have installed with pip previously: !python -m ipykernel install --user --name=my_jupyter_env
  3. Reload the JupyterLab page in your browser
  4. Select the my_jupyter_env kernel from the drop down menu in the top right.