[Use case] Differentiate between point-based and segment-based attributes about sftrack

[basille]

Thanks Jed for this use case! This is actually the basis of the work behind sftraj, and it's going to be tricky. Actually, we've been thinking about it while working on the proposal (see for instance this discussion). At the moment, we're actually thinking of having a dual representation of tracking data (both points and steps), with conversions in between the two — but that's not final, and I'm not sure there's an easy solution which comes with all benefits.

But however we do it, we indeed need to identify the context of ancillary data (associated to point or step). One approach might be to rely (or extend) on the agr attribute of sf objects (see here for more details). I don't know how much this is actually used by sf and other packages though… but could be worth checking and using, instead of reinventing the wheel.

(I'm closing this issue because we have the information necessary to work on this, i.e. requirements, but feel free to add more of your thoughts about it below!)

EDIT: I did not think of implications for plotting, but it's actually very clear that there are strong implications there! No way we can plot point attributes along step/lines…

[jedalong]

Hadn't seen that thread, yes that is a more detailed discussion of some of the issues I am thinking about, wonderful and good luck! I would also re-emphasize the point that to be a truly universal trajectory data model it should be appropriate beyond wildlife tracking applications. One thing is the data sizes (for example lots of human tracking datasets have millions and millions of GPS points from many individuals e.g., geolife), but its true that wildlife ecology may be moving closer to this with ICARUS.

[basille]

Indeed! We are trying to abstract it to a low level, i.e. movement data and trajectories — which are not necessarily wildlife tracking data! There is more and more data from humans… Performance on large datasets (as also seen in this thread) will be key.

[birderboone]

I'm re-opening some of these use-cases so theyre easier to find as people submit new ones.

[basille]

Closing this issue now that sftrack is developed and on CRAN.