nteract / data-explorer

The Data Explorer is nteract's automatic visualization tool.

Home Page:https://data-explorer.nteract.io/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Data Explorer on Google Colab

AksleyRios opened this issue · comments

Is there any way to run de data Explorer on a notebook placed on Google Colab??

Hello @AksleyRios! Thanks for opening this issue.

I'd recommend reaching out to the Google Colab team directly with this feature request. While the nteract organization creates the Data Explorer component, it is up to individual applications to integrate it into their notebook UIs.

@blois of the Google Colab team can point you to the correct place to open this feature request.

commented

Hi @AksleyRios!
It is possible to build Data Explorer in a way that will work in Colab, I put together an example of this a while back in https://colab.research.google.com/gist/blois/9ca9c4556b14f0cdb84d7fe1e5aee691/data-explorer.ipynb.

I haven't worked on this much outside of the initial prototype but I do really like Data Explorer and think it'd be great to get something more officially supported.

For Colab, we try to be as agnostic as possible to extensions like this and implemented a Data Table using a similar approach to what is used for the Data Explorer sample I linked above.

I think the approach I used for Data Explorer in Colab could also be used to enable it in JupyterLab and Jupyter Notebooks without any extensions needed and without any versioning issues- I'd be really interested in seeing a solution that just worked for everyone, everywhere, without any extensions needed.

I think the approach I used for Data Explorer in Colab could also be used to enable it in JupyterLab and Jupyter Notebooks without any extensions needed and without any versioning issues- I'd be really interested in seeing a solution that just worked for everyone, everywhere, without any extensions needed.

I definitely agree with this idea! It'll reduce the friction around adoption of the Data Explorer and make it accessible to the wide majority of people who are using Jupyter classic.

Would you be willing to share how you went about building the prototype so we can make it an official part of the package?

cc: @emeeks Looks like @blois has done some interesting stuff to bring the Data Explorer to Colab. There might be some good ideas there we can generalize and bring to a larger audience.

commented
  1. It's basically just build rules to make a self-contained JS file of Data Explorer, the source code for that is at https://github.com/blois/standalone-dx.

  2. Then the binary is hosted in a versioned directory on a CDN- in this case just using github.io.

  3. The Python code is in the notebook linked above, ideally this would be on pypi.

I'd love to help, just let me know what to do. One thing, I think, would be to genericize the table (grid) view, because so many folks have their own preferred table/grid. Beyond that, I don't know what would best help folks to use it (in nteract and elsewhere).

@emeeks It might be worthwhile to turn the work outlined above into a project for our GSoC participation. The technical scope seems manageable enough a student participant with experience in Python/JS.

I don't think we have bandwith to do the entire work any time soon but tackling it during the summer seems sensible. If this seems good to you, I can add it to our project page.