Include R Images

[santiagobasulto]

Gathering info for now

[jdblischak]

Hi @santiagobasulto! I spoke with @martinzugnoni a few days ago at PyCon 2019 about supporting the R kernel for your awesome notebooks.ai platform. Below are some available resources that might be helpful.

If you're using Docker, a really good resource for R is the rocker project. You could try extending a rocker image by installing Python/Jupyter. And even if you use a different base image, you could learn from their Dockerfiles about how to successfully setup an R environment inside of a Docker container running Debian.

https://github.com/rocker-org/rocker https://hub.docker.com/u/rocker

Another option would be to use conda, and specifically the community channel conda-forge, to install binaries for all the necessary Python and R packages.

https://conda-forge.org/ https://anaconda.org/conda-forge

For example, the following will create a conda environment ready for data science with R and Jupyter:

conda create -n jupyter-r -c conda-forge r-base r-irkernel jupyter r-tidyverse
conda activate jupyter-r

And to combine Docker and conda, I often extend the Docker image continuumio/miniconda3.

[santiagobasulto]

Hey John! Thanks so much for the input. R images are at the top of the list, so we’re going to dedicate some time next week once we’re back from the states. Thanks for the resources, we are currently using Docker images, so it should be similar. Can I ping you once it’s done so you can review it? Is there any other 3rd party libraries that are needed m?

On Thu, May 9, 2019 at 12:13 John Blischak notifications@github.com wrote:

Hi @santiagobasulto https://github.com/santiagobasulto! I spoke with @martinzugnoni https://github.com/martinzugnoni a few days ago at PyCon 2019 about supporting the R kernel for your awesome notebooks.ai platform. Below are some available resources that might be helpful.

If you're using Docker, a really good resource for R is the rocker project. You could try extending a rocker image by installing Python/Jupyter. And even if you use a different base image, you could learn from their Dockerfiles about how to successfully setup an R environment inside of a Docker container running Debian.

https://github.com/rocker-org/rocker https://hub.docker.com/u/rocker

Another option would be to use conda, and specifically the community channel conda-forge, to install binaries for all the necessary Python and R packages.

https://conda-forge.org/ https://anaconda.org/conda-forge

For example, the following will create a conda environment ready for data science with R and Jupyter:

conda create -n jupyter-r -c conda-forge r-base r-irkernel jupyter r-tidyverse conda activate jupyter-r

And to combine Docker and conda, I often extend the Docker image continuumio/miniconda3 https://hub.docker.com/r/continuumio/miniconda3.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/notebooks-ai/notebooks-help/issues/17#issuecomment-490969334, or mute the thread https://github.com/notifications/unsubscribe-auth/AAGU62F5QVMTFFET3JQLVALPUREULANCNFSM4HLNNYKQ .

-- Santiago Basulto.- Up!

[jdblischak]

Can I ping you once it’s done so you can review it?

Yep! I should be able to review it.

Is there any other 3rd party libraries that are needed m?

Do you mean which other R packages would be useful to pre-install for in the images for data science users? Some that come to mind are caret and data.table. If you're feeling ambitious, you could try to include as many packages as you can from the Machine Learning Task View. The caret package suggests many machine learning packages (caret provides a common API to many methods, similar to scikit-learn), so a convenient trick is to install all of these by running this one line:

install.packages("caret", dependencies = TRUE)

[santiagobasulto]

Awesome, thanks @jdblischak!