Bubble: Display order according bubble size

[jq2dj]

Feature Proposal

The bubbles are drawn according the order of the datasets. Bubbles of dataset no. 1 are drawn in front of the bubbles of dataset no. 2. If two bubbles are at the same coordinate, the bubble of dataset no. 1, depending on its size, might covers the bubble of dataset no. 2 completely.

I would like an option for the bubble chart, to change this behavior and draw the bubbles in order of their size. If for a given point two bubbles have to be drawn, the smallest bubble is drawn in front, the greatest bubble is drawn in the back. This ensures that in most cases more bubbles are partially visible (if two bubbles have the same size one bubble is still covered).

Feature Use Case

We use the bubble chart with a very small range of ticks on the x and y axis and multiple datasets. The aim is to compare these datasets against each other. Many, if not all of the bubbles lie an the same coordinate and cover each other. The order of the datasets is secondary in this case.

Currently our workaround is to make the bubble slightly transparent. But when the greater bubble is in front, it is not possible to detect an click event only on the smaller bubble.

[LeeLenaleee]

Duplicate of #3359, since that one has the implement externally label pretty high chance this wont get into the core lib

[jq2dj]

Oh, I looked only through the open issues and missed #3359, my bad - sorry!

But I want to note that in my case sorting the data array by size is not sufficient, as I want the ordering across all datasets and not only within one dataset.

[kurkle]

@JQDJ how/why is your data split to multiple datasets?

Just asking because it would not be trivial to implement draw order sorting across dataset boundaries in the current code base. So I'd look into combining the datasets, and using other options (colors) to make the required distinction between them.
But I don't know enough about the use case to know if that would be sufficient.

But when the greater bubble is in front, it is not possible to detect an click event only on the smaller bubble.

Most of the interaction modes can return multiple items, when multiple items are intersected or are at equal distance from the pointer. So the click detection should be possible.

[jq2dj]

how/why is your data split to multiple datasets?

We have multiple groups of data. Every group has its own filter to select the data.
E.g.:
group A, people with age 0-30
group B, people with age 31-60
group C, people with age 61-90

Then we compare the groups, resp. some attribute of it, in a chart. So it was just natural to map these groups onto datasets.

I have looked into merging all this data into one dataset. It would be possible to keep track of the bubbles and color bubbles according their group with some additional handling outside of the chart.
Unfortunately for us, this would make the bubble chart different from other charts. For the pie, bar and line chart, the "multiple dataset approach" works just fine. I hoped that I can use the bubble chart the same way as the others.

Most of the interaction modes can return multiple items, when multiple items are intersected or are at equal distance from the pointer. So the click detection should be possible.

Yes, that would technically work but it is not suitable for our use case: we implemented a selection functionality, where one can select multiple bubbles. It is not intuitive to the user to select a bubble which is actually in the background, covered by a larger bubble.

[kurkle]

Sounds like you have a valid use case. A proper solution for this kind of use case would be a full z-index support. I don't think that will be possible to implement in v3, because it would require some breaking changes on how the dataset controllers do the drawing.

As always, this is much easier to implement specifically for the use case, but a lot harder to do in a generic manner.

So going to stamp with the implement externally label.