problem with summarizing SpatialPointsDataFrame

[jebyrnes]

I'm wondering if the move to dplyr 0.7 caused some issues.

library(sp)
library(spdplyr)


data(meuse)

coordinates(meuse) <- ~x+y

meuse %>% 
  group_by(landuse) %>%
  summarize(zinc = mean(cadmium))

throws

Error in arrange_impl(.data, dots) : 
  Evaluation error: Column `branch_`: not found in data.

which seems odd...

[mdsumner]

Ah, no - it's just something I never tested. The logic this sits in is a bit convoluted and in some ways was just me learning how to use dplyr, it could be a lot better but I'd prefer to just do

meuse %>%  as.data.frame() %>%
  group_by(landuse) %>%
  summarize(zinc = mean(cadmium))

it could create a multipoint to make it coherent with what group_by/summarize does to polygons and lines - but I consider this low priority given that using sf will now be much more consistent.

here's an sf approach:

library(sf)

meuse %>% st_as_sf() %>% 
  group_by(landuse) %>%
  summarize(zinc = mean(cadmium))

Happy to help you find a better way rather than use spdplyr, are you happy to ignore this and leave spdplyr behind?

(Thanks for reporting the issue, I still might fix it just for the sake of learning)