[IMPORTANT] System Requirements - R + Python Requirements & Common Issues
The software is cutting-edge and new
The software used in this course is very advanced, cutting-edge, state-of-the-art and... Brand new. Because of this you must follow these instructions to set your system up. Otherwise you may run into errors.
RStudio IDE Version 1.3+ (Required)
This is needed to support running reticulate and Python via RStudio IDE python integration. Download RStudio 1.3+ for FREE here.
R Version 4.0+ (Required)
The course was recorded using R 4.0. I've also built & tested the code on R3.6, which works too. The main issue is that as R Packages become updated, they tend to require newer versions of R.
- Upgrading Mac: On Mac, you can simply download and install the newest version of R. When you restart RStudio, it will use the updated version of R.
- Upgrading Windows: I recommend upgrading R via the installr package. This is how I upgraded from R3.5 to R4.0.2.
Development Versions of R Software Libraries (Required)
As of this writing, the software being used for the course depends on new functionality that is only available in the development versions of modeltime, tune, recipes, and workflows. This is support for parallel processing, which should be released to CRAN soon. But until then, if we want to parallel process, we need to use the development versions of the software.
First, update all of your packages to the latest CRAN versions. This should prevent the need to install updates during development package installation. Make sure you have no errors and all packages are successfully updated.
Next, install the development versions of the following R packages:
# Install Tidymodels Development Versions: # ---------------------------------------- devtools::install_github("tidymodels/tune") devtools::install_github("tidymodels/recipes") devtools::install_github("tidymodels/workflows") devtools::install_github("tidymodels/parsnip") # Modeltime & Timetk Development Versions # ---------------------------------------- devtools::install_github("business-science/modeltime") devtools::install_github("business-science/timetk")
Testing Python Integration for the Course (Optional)
If you'd like to test your setup for Python integration, I recommend following my "How to Install TensorFlow 2.0 in R in 5 minutes" article. While we won't be using TensorFlow in this course, it will be a good test to make sure you can get Python up and running.
Mac & Windows (Common Setup Issues)
Both MacOS Catalina & Windows 10 were tested. Either will work.
- XQuartz. The only thing I can think of is that Mac users need to have XQuartz installed. This sometimes gives students a hangup, but in most cases it's already installed when you install R and RStudio.
- RTools: You'll need to have RTools to run any libraries that use C++.
- Python Integration: Windows Python integration may have a few additional steps including modifying the PATH to get the Python-Reticulate setup working using Conda. We'll go through this in Part 3 - Time Series Deep Learning of the course.