explorer VS pyodide

The Explorer library [0] in Elixir uses Polars underneath it.

[0] https://github.com/elixir-explorer/explorer

To ensure you do not miss this: LiveBook comes with a Vega Lite integration (https://livebook.dev/integrations -> https://livebook.dev/integrations/vega-lite/), which means you get access to a lot of visualisations out of the box, should you need that (https://vega.github.io/vega-lite/).

In the same "standing on giant's shoulders" stance, you can use Explorer (see example LiveBook at https://github.com/elixir-explorer/explorer/blob/main/notebo...), which leverages Polars (https://www.pola.rs), a very fast DataFrame library and now a company (https://www.pola.rs/posts/company-announcement/) with 4M$ seed.

Already exists. Check out https://github.com/elixir-nx/explorer which provides a tidyverse-like API in Elixir using polars as the back end.

José from the Livebook team. I don't think I can make a pitch because I have limited Python/R experience to use as reference.

My suggestion is for you to give it a try for a day or two and see what you think. I am pretty sure you will find weak spots and I would be very happy to hear any feedback you may have. You can find my email on my GitHub profile (same username).

In general we have grown a lot since the Numerical Elixir effort started two years ago. Here are the main building blocks:

* Nx (https://github.com/elixir-nx/nx/tree/main/nx#readme): equivalent to Numpy, deeply inspired by JAX. Runs on both CPU and GPU via Google XLA (also used by JAX/Tensorflow) and supports tensor serving out of the box

* Axon (https://github.com/elixir-nx/axon): Nx-powered neural networks

* Bumblebee (https://github.com/elixir-nx/bumblebee): Equivalent to HuggingFace Transformers. We have implemented several models and that's what powers the Machine Learning integration in Livebook (see the announcement for more info: https://news.livebook.dev/announcing-bumblebee-gpt2-stable-d...)

* Explorer (https://github.com/elixir-nx/explorer): Series and DataFrames, as per this thread.

* Scholar (https://github.com/elixir-nx/scholar): Nx-based traditional Machine Learning. This one is the most recent effort of them all. We are treading the same path as scikit-learn but quite early on. However, because we are built on Nx, everything is derivable, GPU-ready, distributable, etc.

Regarding visualization, we have "smart cells" for VegaLite and MapLibre, similar to how we did "Data Transformations" in the video above. They help you get started with your visualizations and you can jump deep into the code if necessary.

I hope this helps!

There's a package in the Nx ecosystem called Explorer (https://github.com/elixir-nx/explorer). It uses bindings for the rust library, polars, which is much more betterer than Pandas.

FWIW around April this year I started using erlport with python polars in a production ETL app because explorer didn't have the features I needed at the time.

Machine learning is rapidly expanding within the Elixir ecosystem, with tools such as Nx, Axon, and Explorer being used both by individuals and companies such as Amplified, as mentioned above.

We implemented an in-browser Python editor/interpreter built on Pyodide over at Comet (our users are data scientists who need to build custom visualizations quite often, and the most familiar language for most of them is Python).

One of the issues you'll run into is that Pyodide only works by default with packages that have pure Python wheels available. The team has developed support for some libraries with C dependencies (like scikit-learn, I believe), but frameworks like PyTorch are particularly thorny (see this issue: https://github.com/pyodide/pyodide/issues/1625 )

We ended up rolling out a new version of our Python visualizations that runs off-browser, in order to support enough libraries/get the performance we need: https://www.comet.com/docs/v2/guides/comet-ui/experiment-man...

Thank you! Yes, one of the items in the Roadmap is support for Pyodide (https://github.com/pyodide/pyodide) for running in-browser python on the results of each of the code blocks! This should allow most ML libs to be usable in-browser! This is pretty high-up on our priority list.

If I understand correctly, WASM only makes sense for compiled languages, you can run the python interpreter in WASM of course[1], but that will be at a significant performance disadvantage to the native javascript interpreter, and it's also something that has to be loaded every time you load the website.

[1]: https://github.com/pyodide/pyodide

If you absolutely need something comparable to Sympy, then one option might be to figure out how to best call Sympy from Rust. e.g. - RustPython, although it seems like Sympy isn't supported yet - Pyodide, and figuring out how to run it outside of a web browser. Probably also not very easy. - PyPy, and having a pretty simple Python binary for every platform - ...

For running programming languages other than JavaScript in the browser there is Emscripten and WebAssembly. There is v86, where a Linux build is compiled to WASM. Folks have written QuickJS into a Linux build compiled to WASM, Node.js into the Linux buildroot https://github.com/cemalgnlts/now, so Python or CPython can be written to the image and loaded into the browser as WASM as well https://github.com/pyodide/pyodide.

There's a Python runtime that runs on WebAssembly (https://github.com/pyodide/pyodide). I have no idea what it's like, I've never used it.

Still in a quest to provide some tooling to quickly compose documentation websites: https://github.com/synw/docdundee . As I have tons of libs to document and was tired of managing restructured language for readthedocs I started with this, and now it has executable Python examples in the frontend via a Pyodide wrapper composable: usePython

This project's brand new and a lil hacky, but I've already reached out to the scikit-learn team, and they recommended that I experiment with using Pyodide as an alternative backend for the Python bridge.