sample, working jupyter notebooks and R code to host

[emiliom]

We now have two code/script directories ready:

• https://github.com/ioos/BioData-Training-Workshop/tree/master/notebooks
• https://github.com/ioos/BioData-Training-Workshop/tree/master/rcode

We already have 3 jupyter notebooks in the notebooks directory (thanks to @ocepaf for uploading his erddapy notebook!). I mentioned I loaded a relevant "R vignette" from Roy Mendelsohn and Scott Chamberlain into the rcodedirectory, but we'll need an R person to test it, or even tell us/me if that's something that's actually meant to be run as is (I have no clue).

I'm starting this issue to discuss and track the loading of additional samples, plus testing them to ensure they actually run with our setups.

Next steps:

• I just noticed the R obistools test jupyter notebook in the tests directory, https://github.com/ioos/BioData-Training-Workshop/blob/master/tests/test_obistools.ipynb. Looks cool! Let's copy it into the notebooks folder and tweak it if needed!
• @lsetiawan Please go ahead and test the jupyter notebooks on JupyterHub and locally. Tweak as needed to make them work, with the caveat that I don't expect you to be able to fully fix the notebooks that are actually in R
• @bbest can you help us with the R side of things?? That includes loading a couple of R scripts/vignettes/notebooks you think would be appropriate and of interest

[lsetiawan]

• I was able to run all of 2017-08-01-xtractoR.ipynb with no problem.
• For the 2017-11-30-rerddap.ipynb I get a ModuleNotFound error for rpy2 I tried installing rpy2 and re-run the notebook, but now I get ImportError: libicuuc.so.54: cannot open shared object file: No such file or directory.
• Then the last notebook, filipe_fernandes_intro_errdapy.ipynb runs fine after I installed cartopy.

[bbest]

Hi @emiliom,

Thank you for getting this organized. I'll do a deep dive on these tomorrow.

[emiliom]

Thank you for getting this organized. I'll do a deep dive on these tomorrow.

Thanks @bbest. If you can focus on the non-notebook R scripts/vignettes/stuff, that'd be optimal. The rest of us are already handling the Jupyter-notebook sample code.

[emiliom]

@lsetiawan, with PR #20 merged, have you tried 2017-11-30-rerddap.ipynb again?

(Just checking while I wait for some other stuff to finish :smile_cat:)

[lsetiawan]

The kernel keeps dying on me now when I try to run 2017-11-30-rerddap.ipynb

[emiliom]

The kernel keeps dying on me now when I try to run 2017-11-30-rerddap.ipynb

Oh well. Maybe @ocefpaf can look into it on Wednesday once he's here in Seattle, since he had a role in creating the notebook. Right now I think he's up in the air somewhere.

Thanks for testing.

[bbest]

Per Roy Mendel's https://github.com/ioos/BioData-Training-Workshop/issues/13#issuecomment-363248885, these notebooks are worth checking out for using xtractomatic, rerddap and rerddapXtracto:

https://rmendels.github.io

[emiliom]

@bbest, can you clarify (for me at least!) what is a "notebook" in a non-Jupyter R context? I see in Roy's page he points to several "vignette RNotebooks", and see that the downloadable Rmd files look like mostly Markdown with a bit of magic thrown in. Are these "vignette RNotebooks" Rmd files executable (say, in Rstudio) the same way a Jupyter notebook ipynb file is in Jupyter?

[bbest]

Hi @emiliom, the R notebooks are newer than typical Rmarkdown documents and meant to be similar in functionality as Python notebooks which both contain the reproducible code and rendered output. To be honest, I haven't been using them yet, but here's the minor extra functionality I gleaned...

1. Source Rmd is embedded. Notice in the notebooks at https://rmendels.github.io, in the upper right you can "Download Rmd" which is embedded inside the nb.html, so you don't need to distribute two seperate files as with typical Rmarkdown (ie source *.Rmd and rendered *.html).

2. Preview is instantaneously updated. There is also a slightly different workflow allowing for instant preview of html result in the Viewer pane as you execute lines (or chunks) of code, whereas with typical Rmarkdown you need to Knit (ie render()) the whole document in order to see the updated html result in the Viewer.

   

Format Summary. Many other extremely useful Rmarkdown output formats (PDF, Word, beamer, Websites...) work in the more typical fashion of needing to Knit first and don't have the source embedded.

• R notebooks: *.Rmd rendered to *.nb.html using output: html_notebook in front matter
• Rmarkdown to html: *.Rmd rendered to *.html using output: html_document in front matter

References:

• Rmarkdown output formats
  • HTML Documents
  • R Notebooks - Saving and Sharing
  • R Notebook HTML Format
  • R Notebook Workflows (rendered presentation)

[emiliom]

Wow, thanks for that rich background @bbest !

So, what I'm hearing (and also saw at http://rmarkdown.rstudio.com/r_notebooks.html) is that an *.Rmd document that was created as an R notebook can in fact be opened and executed in RStudio, much like an *.ipynb jupyter notebook document in Jupyter. Great.