Re:Why Jupyter is data scientists’ computational notebook of...
by azag0 on Tuesday October 30, 2018 @02:51PM
(#18340116)
I love Jupyter Notebook for experimenting and rapid creation of reports, but dislike it for not being able to use my editor and for intermingling inputs and outputs in a single file. So I'm working on an alternative frontend to Jupyter kernels, which is heavily inspired by KnitR: https://github.com/azag0/knitj It is still being developed, but it's functional and I use it every day.
[ Parent
| on HN ]
Re:Why Jupyter is data scientists’ computational notebook of...
by y4mi on Tuesday October 30, 2018 @03:55PM
(#18340723)
> for not being able to use my editor
At least Atom has good integration with hydrogen
Re:Why Jupyter is data scientists’ computational notebook of...
by bodkan on Tuesday October 30, 2018 @07:30PM
(#18342137)
This looks incredible! Does your project already support other language kernels than the Python kernel?
I use R for 90% of my work, but most of it has been happening in Jupyter notebooks (which I'm not a huge fan of, despite practically living in them for the past 4 years of my life).
Thanks for sharing!
Re:Why Jupyter is data scientists’ computational notebook of...
by azag0 on Wednesday October 31, 2018 @02:08AM
(#18343670)
I have not tested it with anything else than the Python kernel, but it uses Jupyter Client to communicate with the kernel, which is kernel agnostic. So you should be able to do just “knitj -k <kernel name> ...”.
Re:Why Jupyter is data scientists’ computational notebook of...
by infinite8s on Wednesday October 31, 2018 @10:35AM
(#18346006)
If you use R so much why do t you use RStudio instead?
Re:Why Jupyter is data scientists’ computational notebook of...
by huac on Tuesday October 30, 2018 @03:43PM
(#18340604)
your project also seems to play much more nicely with git!
Re:Why Jupyter is data scientists’ computational notebook of...
by azag0 on Tuesday October 30, 2018 @03:50PM
(#18340673)
True. Actually that's what I meant by "intermingling inputs and outputs". KnitJ still shows both code and its output in the rendered HTML, but unlike in Jupyter Notebook, the code is stored and edited separately in a single source file.
Re:Why Jupyter is data scientists’ computational notebook of...
by mikepurvis on Tuesday October 30, 2018 @03:57PM
(#18340750)
That's my biggest frustration by far with trying to use Jupyter for anything Serious Business. Will definitely check out your project; thanks for sharing.