Scale domain vs filter selection: automatic axis scaling

[aldanor]

In Scale Domains docs it is noted:

An alternate way to construct this technique would be to filter out
the input data to the top (detail) view like so:

{
  "vconcat": [{
    "transform": [{"filter": {"selection": "brush"}}],
    ...
  }]
}

Which is indeed almost the same (although filter method being much slower, as noted in the docs), except for one difference:

• With filter-selection method (demo), the y-axis of the upper chart will be automatically zoomed in to the selected points. This is pretty neat, especially if you have large amount of points.
• With scale-domain method (demo), the y-axis remains frozen as you move the selection around.

The question: is it possible to have the y-axis automatically zoom in to the selected points as you move the selection, with "scale domain" method (same as it does with filter-selection method)?

// I've checked all scale-domain issues here, on altair repo and SO and couldn't find anything related; apologies if it's something obvious.

[domoritz]

Please use stack overflow or slack for questions. We reserve the issue tracker for bugs and feature requests (including doc requests).

[aldanor]

@domoritz I would've posted a feature request if someone could confirm this is currently impossible, hence the question (apologies for hassle).

I've posted an SO question - hoping someone could answer that.

[domoritz]

If you want to filter the data to the selected points or set the scale domain to only include some filtered data, you need to use the filter method.

[aldanor]

@domoritz Thanks. The problem is, scale domain method with binding to x-axis works surprisingly fine if there's around 100K points in the dataset, whereas filter method almost dies.

I was wondering, if scale domain method is so efficient (because it doesn't have to scan through all data points along the x axis), is there a faster way to auto-scale the resulting plot along the y-axis? If this was possible, it would make vega-lite miles ahead all of the existing tools.

Alternatively, maybe it was possible to show the full dataset (with scale domain), but auto-scale it along y-axis using heavily resampled dataset? (so that it can compute y-min and y-max much faster on a smaller number of points). Not entirely sure that's possible in vega-lite.

// I was even thinking, it might be even fine to have to pre-compute some mins/maxes for a quantized matrix of (start, end) selection ranges and pass them to vega-lite as a separate dataset, but not sure if it's doable either.

[domoritz]

Can you find out what's slow? I'd rather improve the performance of filtering than support two ways to achieve the same thing. That would be fantastic. Feel free to open a new issue with the goal of performance improvements for your use case. Thanks!