How to count the number of points in a polygon by a grouping variable in R?

Question

I am interested in calculating the proportion of GPS points that fall within protected areas. I can do this with my full data set with the code I show below.

However, I would like to group this data by the id column and get a measure of overlap for A and B separately.

Thanks

# load packages
library(wdpar)
library(dplyr)
library(sf)

# download protected area data 
# (excluding areas represented as point localities)
sa_raw_pa_data <- wdpa_fetch(c("South Africa"))
sa_raw_pa_data 

# reproject data
sa_pa_data <- st_transform(sa_raw_pa_data, 4326)

# plot it
# plot(sa_pa_data)

# create some random data
set.seed(123)
d = data.frame(long=runif(100,16,38), lat=runif(100,-47,-22))
d$id <- c(rep("A", 50), rep("B", 50))

# measure the overlap
pts <- st_as_sf (d, coords = c ("long", "lat"), crs = 4326)
res <- st_intersects (pts, sa_pa_data)

length (unlist (res)) / nrow (pts) # fraction of intersecting points

Answer 1

You could split your points into a list where each element is a group (A, B). Then you can use lapply to iterate over each group.

point_list <- split(pts, pts$id)  # split points into list by id

res <- lapply(point_list, function(x) 
   length(unlist(st_intersects(x, sa_pa_data))) / nrow(x)
)

unlist(res)
   A    B 
0.06 0.02