Combine polygons by attribute

[Nowosad]

related to #275

I've combined polygons by attribute using group_by() and summarise(). It leaves some lines, so I've tried @edzer tip about using st_set_precision(). However, every value I've tested (100 - 1000000000) keeps some "artefacts". What's the best way to deal with that kind of problems?

# devtools::install_github("nowosad/spData")
library(spData)
library(sf)
library(tidyverse)

world_continents1 = world %>% 
  group_by(continent) %>% 
  summarise(pop = sum(pop, na.rm = TRUE), country_n = n())
plot(world_continents1[1]) 

world_continents2 = world %>% 
  st_set_precision(10000) %>% 
  group_by(continent) %>% 
  summarise(pop = sum(pop, na.rm = TRUE), country_n = n())
plot(world_continents2[1]) 

[edzer]

Which lines do you see? You're sure these are not holes?

[Nowosad]

Yes, I'm certain.

Data before group_by() and summarise() - no holes:

plot(world[0])

Data after group_by() and summarise(), different value of precision - "artefacts" in a different places (e.g. North America):

world_continents3 = world %>% 
  st_set_precision(100) %>% 
  group_by(continent) %>% 
  summarise(pop = sum(pop, na.rm = TRUE), country_n = n())
plot(world_continents3[1]) 

[edzer]

None of the geometries you get are actual lines, all have dimension 2 and positive area:

> world_continents2[1] %>% st_area() %>% units::set_units(km^2)
Units: km^2
[1] 29946212.22 12335958.52 31252444.82 23065279.95 24484309.31  8504478.20
[7]    11602.57 17762597.45
> world_continents2[1] %>% st_dimension()
[1] 2 2 2 2 2 2 2 2

what is of course possible is that there's no man's land between polygons. Africa seems to have 4 holes:

> world_continents2[1] %>% st_cast("MULTIPOLYGON") %>% st_geometry() %>% sapply(lengths)
[[1]]
[1] 1 5

[[2]]
[1] 1 1 1 1 1 1 1 1

[[3]]
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1

[[4]]
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 1

[[5]]
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[39] 1 1 1 1 1 1 1 1 1

[[6]]
 [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

[[7]]
[1] 1

[[8]]
[1] 1 1 1

Maybe use mapview to explore them, and mapedit to remove them?

[tim-salabim]

Thanks for the use-case :-)

[Nowosad]

@edzer Thanks for the investigation. I'm just still unsure why the number and location of these holes changes with different values of a "precision".

@tim-salabim I'll definitely try to use mapedit.

[edzer]

@Nowosad see also ?st_as_binary for more docs, also try precision values smaller than 0; create a toy example to understand how it works.