Closed njtierney closed 6 years ago
See #51 and geodist
. For this kind of stuff, the default mapbox cheap ruler distance is plenty adequate - a pretty constant 0.16% error compared to 40-50 times slower sf
-type distances. (Sorry, just trawling through your issues here; I'll stop now and let you close some before I have me another look.)
Super keen to implement geodist
!
I'm going to dust off and arrange the issues today, but don't apologize for trawling through the issues, I'm so stoked that you are interested in this!
Closing this issue as the distance calculation will in the future be done with geodist
now - see #50
The fields R package implements a few measurements of measuring distances, see
rdist.earth
andfields.rdist.near
.It would be worthwhile to evaluate how
maxcovr
is different to these functions (and others), does it perform better? I think currently it is nice to have something that is more 'tidy' and doesn't require longitude and latitude to be in matrix pairs.Perhaps in the future it might be worthwhile to extract out the distance helper functions in maxcor into their own package that is slim, fast, and tidy. Although, I get the sense that perhaps
widyr
might achieve this, or that there might be some other packages that do this.