Install Rkernel For Jupyter

1 minute read

Published:

Install R kernel for Jupyter notebook

This instruction works on Linux and macos platform. It is mostly borrowed from this post and post, and has been tested on Linux machine.

Create conda env and install r-irkernel

conda create -n r-kernel
# activate it
conda activate r-kernel
(r-kernel)$ conda install r-recommended r-irkernel jupyter

Install a specific version of R

# create conda env
conda create --name r_3.4.3
conda activate r_3.4.3
# search and install a different r version
(r_3.4.3) $ conda search r-base
(r_3.4.3) $ conda install -c r r=3.4.3
# after successful installation. check the R version
(r_3.4.3) $ which R
(r_3.4.3) $ R
# install irkernel and jupyter 
conda istall -c r r-irkernel jupyter

Add r kernel to jupyter kernel spec [important!]

(r-kernel)$ R -e 'IRkernel::installspec()'

(optional) check kernel spec

  • See a list of kernel spec files
$ jupyter kernelspec list
Available kernels:
  exodus             /global/u2/p/pshuai/.local/share/jupyter/kernels/exodus
  geo                /global/u2/p/pshuai/.local/share/jupyter/kernels/geo
  ir                 /global/u2/p/pshuai/.local/share/jupyter/kernels/ir # make sure this exists!
  • Change jupyter display kernel name
$ cd /global/u2/p/pshuai/.local/share/jupyter/kernels/ir
$ vi kernel.json

# change the display_name below
{
  "argv": ["/global/homes/p/pshuai/.conda/envs/r-kernel/lib/R/bin/R", "--slave", "-e", "IRkernel::main()", "--args", "{connection_file}"],
  "display_name": "iR", 
  "language": "R"
}

Install R packages

  • on command line
R -e 'install.packages("rgdal", repos = "https://cloud.r-project.org")' # avoid CRAN selection
  • inside R

Open R from the conda env, and use the install.packages("<package_name>")