How to set a geometry to NA (empty) for some features of an sf dataframe in R

Question

I would like to set to empty the geometries of some features of my sf object given a condition. Here is my code:

my_points %<>% mutate(geometry = if_else(district_x != district_y, st_sfc(rep(st_point(), nrow(.))), geometry))

Basically, my dataframe contains two text columns named district_x and district_y respectively and a POINT geometry (sfc) column named geometry. Everytime the value in district_x is different from the value in district_y, I want to set the geometry of that feature to Empty.

With this code above, I am getting the following error:

Error in `stopifnot()`:
ℹ In argument: `geometry = if_else(...)`.
Caused by error in `vapply()`:
! values must be length 3,
 but FUN(X[[1]]) result is length 1

Any hints?

Answer 1

I'd set the geometry elements to st_empty() using a logical subselection. Here's a reproducible example that only uses the sf package - you seem to have some other unstated packages attached.

First lets make a sample data set:

d = st_as_sf(
data.frame(
district_x=c(1,2,3,4),district_y=c(1,5,3,2),
x=c(1,2,3,4),y=c(1,2,3,4)),
coords=c("x","y"))

Which is this:

> d
Simple feature collection with 4 features and 2 fields
Geometry type: POINT
Dimension:     XY
Bounding box:  xmin: 1 ymin: 1 xmax: 4 ymax: 4
CRS:           NA
  district_x district_y    geometry
1          1          1 POINT (1 1)
2          2          5 POINT (2 2)
3          3          3 POINT (3 3)
4          4          2 POINT (4 4)

Which rows are we interested in? Let's compare columns:

> d$district_x == d$district_y
[1]  TRUE FALSE  TRUE FALSE

Change those elements:

d$geometry[d$district_x == d$district_y] = st_point()

And now they are changed:

> d
Simple feature collection with 4 features and 2 fields (with 2 geometries empty)
Geometry type: POINT
Dimension:     XY
Bounding box:  xmin: 2 ymin: 2 xmax: 4 ymax: 4
CRS:           NA
  district_x district_y    geometry
1          1          1 POINT EMPTY
2          2          5 POINT (2 2)
3          3          3 POINT EMPTY
4          4          2 POINT (4 4)

Actually I'm not sure if I've changed the ones you want or the ones you didn't, but flip the test if so. Like:

> # reset `d`:
> d = st_as_sf(data.frame(district_x=c(1,2,3,4),district_y=c(1,5,3,2),x=c(1,2,3,4),y=c(1,2,3,4)), coords=c("x","y"))
> # test and set
> d$geometry[d$district_x != d$district_y] = st_point()
> # look
> d
Simple feature collection with 4 features and 2 fields (with 2 geometries empty)
Geometry type: POINT
Dimension:     XY
Bounding box:  xmin: 1 ymin: 1 xmax: 3 ymax: 3
CRS:           NA
  district_x district_y    geometry
1          1          1 POINT (1 1)
2          2          5 POINT EMPTY
3          3          3 POINT (3 3)
4          4          2 POINT EMPTY

The principle is: Create a vector of TRUE/FALSE for each row, set the geometry for the TRUE elements.

Strictly you should probably use st_geometry to get the geometry rather than $geometry since it might not be named that way in another data set:

> st_geometry(d)[d$district_x != d$district_y] = st_point()

Note this doesn't need any other packages which seem to be confusing you. Using simple constructions that can be easily taken apart and inspected will advance your learning.