One of my current programming hobbies (long since "in the works" but really just "in the brain") is collecting and refactoring all of my R code and writing the same code in Python. The main purpose of refactoring my R code is to introduce myself to new libraries and functions in R, while also cleaning up code I wrote at the beginning of my R coding career. By rewriting these R scripts in Python, I can interactively and practically learn a new language.

I'll be updating this post or creating new posts as progress is made, but for now the R code has been collected as is viewable in GitHub. The plan is to create a new repo for refactored R scripts and yet another repo for the corresponding Python scripts. I'll likely use IPython/Jupyter for the Python scripts to track my notes as I learn. I don't know if R code can be embedded in Jupyter notebooks, but if it can then perhaps I can save a step.

Thanks for reading. -Edward

UPDATE: Yep, R can be run in Jupyter. Fun. There are at least 2 ways to run R in Jupyter:

1. Run R "natively" using the R Kernal in Jupyter
2. Run R within the Python kernel using the rpy2 package

R Kernel

To run R natively in Jupyter, IRkernel needs to be installed. I had a whale of a time installing the required dependency packages in R due to missing libraries (most notably libcurl and libcurl-openssl-devel). Within openSUSE Leap 42.1 I wasn't even able to locate the latter package within Yast. A Google search for the rpm file did the trick though. Oh, and gcc-c++ was needed as well for compilation. Also, install_github didn't work at somet point. Had to use install_git instead.

R within Python Kernel

To do this, the rpy2 package needs to be installed. I had all sorts of compilation issues using pip to install this, so I installed using the package manager (zypper in SUSE).

Once installed, open up a Jupyter notebook and run:

%load_ext rpy2.ipython

That's it! R code can then be run using the R-magic command: prepend \%R before code.

UPDATE 2:

I've given this more thought, and I believe a better course of action will be to choose a relevant data science technique and code some real data analysis in both R and Python. I can grab some fun data easily from an open source repository. I suspect most of the recoding I was planning to do will get covered in this practice anyway. Next on the agenda then is to identify the techniques I'm interested in coding, prioritize, find data, and begin. Next update will list the techniques.