Numeric scale could be set to aggregate by a number of rounded units just like baseUnit=fit/maxDateGroups does.
Does Kendo plan to provide this kind of functionality?
We would need to draw area charts on numeric scale and use shared tooltips.
A side note: "scatter" option is currently missing from series type documentation:
http://docs.kendoui.com/api/dataviz/chart#configuration-series.type
V
3 Answers, 1 is accepted
If I understand correctly, you'd like to aggregate points that fall within a specified range? Say, all points with X values from 0 to 1 should be represented with a single X/Y value?
I'm not sure I've seen such functionality before. Please, feel free to cast your vote for such feature on UserVoice
Thanks for spotting the omission in the documentation. We'll fix it right away.
Regards,T. Tsonev
Telerik
Rank 1
Hi, Tsonev
The idea maybe goes too far, but let me get back to practical needs. Please follow and see if there's anything to pick up :)
As mentioned we have floating-point data with numeric-scale x-axis.
User needs the possibility to see series (y-axis) closest value while moving cursor anywhere in plot area (as kendoui shared tooltip does). Currently it's difficult to "hunt" for a specific point in scatter chart. Would that be possible?
Secondly the shape of the curve is important to the user. That would be much better visible with area charts (some graph engines even do thin columns).
It would all be possible if I converted my data to (say 30) categories of rounded value-pairs. So my next thought was: it's exactly what kendoui date series already does (with maxdategroups). On numeric data the feature could be called e.g. "trend values".
To sum it up it would be 3 ideas:
1) Shared tooltip for scatter chart (showing closest y-value)
2) scatterArea series option
3) "trendvalues" option for scatter data as described above
Do some of those make sense to you?
Thanks.
V
Thank you for taking the time to share your suggestions.
Having a shared tooltip in the scatter charts is a good usability feature. We need to keep an eye on the performance, especially with large data sets. Partitioning the points in a k-d tree will probably help here.
Scatter area series should be fairly simple to add. As long as we make it clear that point order is important, the surprise levels should be fairly low.
I find the "trend values" idea very intuitive. We can have a separate type of series with a configurable "resolution". That'd be much more intuitive than hunting for the right property to set.
I've noted these ideas and I hope they'll make it on our roadmap soon. As a token of gratitude for your involvement your Telerik points have been updated.
Regards,
T. Tsonev
Telerik