Document how VegaFusion solves real problems

[MarcSkovMadsen]

After having read the material surrounding vegafusion I think the value proposition could be made clearer. I am not an experienced Altair/ Vega user and read it from that perspective. So starting out with "small" examples that already work in Altair does not really show what difference vegafusion provides.

Altair Documents how to work with big datasets

According to https://altair-viz.github.io/user_guide/faq.html#maxrowserror-how-can-i-plot-large-datasets you can work with large datasets if you just follow some best practices.

Why and how does VegaFusion provide something better?

User on twitter says he has worked with altair and 20 mill rows without problems.

https://twitter.com/ArnaudovKrum/status/1485971877029888007?s=20&t=GXoAzB20cc-8Z7mlA2ZwMg

Alternatives

There are Python alternatives like Datashader+Holoviews and Vaex (?). How does vegafusion compare to those?

Users with performance problems

I believe a google search will show lots of users having hit the max limit of what Altair/ Vega can do. Can Vegafusion solve those issues?

New, real world use cases

Examplify what Vegafusion can do via some new big data use case that solves a real problem or provides real insights.

[jonmmease]

Thanks for the documentation notes @MarcSkovMadsen, I think these are all good suggestions.

[MarcSkovMadsen]

I would also like to know

• Is it right to think that Altair/ Vega already supports 80% of users/ use cases.
• Vegafusion can bring that from 80% to maybe 95%?
• HoloViews+Datashader can bring that from 80% to maybe 99% as they can scale out to cluster?

Or how do you see things?

[MarcSkovMadsen]

Is it right to think that the potential adoption of VegaFusion over DataShader+HoloViews will depend on the fact that Altair/ Vega is seen as a more "modern", "easy to use", "beautiful" plotting framework that for Vega is highly popular across programming languages? Not so much on features?

[MarcSkovMadsen]

Bokeh provides a lot of "scientific" tools. How do you see the use of VegaFusion in science? Can it compete? And can it at all scale to big data scientific use cases?

Or is VegaFusion more appealing to enterprise big data use cases and maybe framework/ tool builders across programming languages?

[MarcSkovMadsen]

If one day HoloViews added Altair as a plotting backend. What would that mean for VegaFusion?