Hi. I’m Sharon Machlis at IDG Communications, here with a special episode of Do More With R: How to set up your system for Python.
It’s “special” because I’ve got a guest today: Serdar Yegulalp, InfoWorld’s Python expert and host of the InfoWorld Dev with Serdar video series. He’s here to help answer questions we R users might have when installing and configuring Python for RStudio.
Let’s get to it!
If you are working locally, the reticulate R package has an easy Python install command: install_miniconda().
If you run that you should see a response something like this. R is running commands to install Python, install some Python packages, and create a virtual environment. But it can be hard to understand what’s going on here.
Instead, since we’ve got Serdar here to help, I’d like to go through a manual version: RStudio’s suggested workflow, step by step.
Step 1, not surprisingly, is to install Python. But we’ve got choices! Serdar, would you recommend downloading from python.org or Anaconda? Another question I often run into for Python in general: Should I use Python version two-dot-X or three-dot-X?
RStudio says we need the virtualenv Python package. That requires Python’s pip install command, which I’ll run in a terminal window. While I run pip install, Serdar, can you tell us why we need virtual environments?
Next, step 2, is to create a Python virtual environment for an RStudio project. I’ll open an R project in RStudio and then create my virtual environment. Again, notice that I’m running that virtualenv command in a terminal window, and not the R console. Serdar, why should we use one virtual environment per project?
Step 3 is to activate my virtual environment with the source command. Anything we need to know about this?
Step 4 is a familiar one: Install packages we want. But these are Python packages. So instead of install.packages() in our R console, we need to run pip install in a terminal window to install Python libraries. Serdar, any other must-have packages you’d recommend for an R user doing data analysis or other common work?
The last step is in R. We need to install the reticulate package if it’s not already on our system, and then load the reticulate package. reticulate was designed to help Python and R interoperate, and it allows for easy data transfer between the two. We also need to set an R environment variable so reticulate knows where python is. I’ll need to restart my R Session for this to take effect. Now, let’s test!
There are several ways to run Python code within RStudio. You can add Python chunks to an R Markdown document. You can run Python code in your R script with reticulate’s py_run_string() function. You can library directly in your R code with reticulate’s import() function, or source a Python script from R with py_run_file(). Or you can run Python the conventional way from a console – including an RStudio console.
There’s a separate Do More With R video on running Python within RStudio you can watch after this one.
That’s it for this episode! Thanks, Serdar, for your Python tips; and thank you for watching! For more R tips, head to the Do More With R page at bit-dot-l-y slash do more with R, all lowercase except for the R.
You can also find the Do More With R playlist on YouTube’s IDG Tech Talk channel -- where you can subscribe so you never miss an episode. And for Python tips, check out the Dev with Serdar series in the same channel. Hope to see you next time. Stay healthy and safe, everyone!