Calling Python from R

I’ve recently been trying to use R to leverage some of the excellent packages developed in this language. But my expertise is in Python and I wanted to avoid having to learn a new language, at least in an early phase when I’m just trying things out.

Enter reticulate, a package which can connect Python and R.

Assuming your R environment is set up properly (I found it quite easy to use anaconda to manage R, by just creating a new environment), all you need to do is install the package


and, after you load it


Python will be readily available:

np = import("numpy")

a = np$linspace(0, 1, 100)
  [1] 0,00000000 0,01010101 ...
 [97] 0,96969697 0,97979798 0,98989899 1,00000000

# could be done in standard R with 
a = seq(from=0, to=1, length.out=100)

stats = import("scipy.stats")
[1] 0,3989423

# could be done in standard R with 
[1] 0,3989423

reticulate works by embedding a Python session within the R session, allowing for this incredibly simple interoperability. Behind the scenes, all (Python/R) data types are being automatically converted to their equivalent (R/Python) types. When values are returned from Python to R they are converted back to R types. Everything seems to just work!

I did notice a performance hit: calling the Python functions was significantly slower than calling the R functions (though I only tested very few cases). This is to be expected, but I don’t think it is a show stopper at all. At the level that I intend to use this package, the time I spend writing code is much more important than the time the code takes to run.

wrap up

A powerful but simple way to call Python from within R. Give a try!

btw: this is #AcaDoWriMo!