Line Chart: More colour series, and daily binning on time series


Our data scientists want to visualise a chart, with up to 10’000 lines, plotting a metric over time (say 10 yrs), with daily binning.

Currently Immerse will only pre-select some of the Colour dimension values (say 5 out of the full 10k), and while we can add additional colours, it only allows maybe 15 extra, and o course we wouldn’t want ot manually add 9000+ entries.

When selecting a date dimension for the X-axis, the finest granularity for binning is month, we’d like to be able to go to daily.


So basically you want to plot 36 500 000 data points on a line chart. While it can probably be done using MapD-Core, I see it more as a challenge on the datavis side than technological (e.g., overplotting, selecting, legend size, color rotation). If the analytics task really forbids smoothing, binning, filtering, subsetting or any reduction, it seems like a fairly specific use case that would require custom tools optimized for the task (e.g., stream loading, canvas rendering, custom color palette, voronoi selection). I would be glad to point you to the right resources to build custom analytics on top of MapD’s open-source components.


@biovisualize this is very true, as of the more general questions of:
(1) time granularity (day, minutes, seconds?). Is higher time granularity going to be allowed?
(2) easier/faster/larger selection of names on the colour dimension. Is this something one can do?



These are good ideas for improving Immerse. It’s not on the roadmap so far but I’m transmitting them to the Product team to get prioritized. One thing we are currently working on is to build more examples of how to work directly with our open source components to build special cases like this. We will find a way to announce it, probably with a blog post. It will also be the subject of a workshop we are giving at OpenVisConf in Paris in May Stay tuned!