Find two points farthest apart using group_by() in Sf - r

Question

Haven't been able to find an exact Q/A to match this problem, though there are several related. Im trying to calculate a distance matrix for all points groups defined by an ID column. Then select the two points that are farthest apart from each group, retaining the original group id. The number of points in each group varies from 2, 4 or 6.
My sf df:
df <- data.frame(x = runif(12), y = runif(12), id = rep(1:3,each = 4)) %>%
st_as_sf(coords = c("x","y"), crs = 27700)
I've tried code such as:
a <- df %>%
group_by(id) %>%
st_distance(.)
Though this just returns a distance matrix of all points.
The below gives me, what I want, though I fear it would be slow on large datasets:
maxMin <- do.call(rbind,lapply(unique(allInts$id), function(x) {
df <- allInts %>% filter(id == x)
d <- st_distance(df)
df %>% slice(unique(as.vector(which(d == max(d),arr.ind=T))))
}))

Answer 1

You can use dplyr::group_split to split your data frame into a list per group. You can then apply whatever function you want to that list using map/lapply.
Script below keeps the 2 points which are furthest apart in each group.
library(sf)
library(tidyverse)
# dummy data
data <- data.frame(x = runif(12), y = runif(12), id = rep(1:3,each = 4)) %>%
st_as_sf(coords = c("x","y"), crs = 27700)
# split it into a list per ID
data_group <- data %>%
group_by(id) %>%
group_split()
#apply a function to each list
distance_per_group <- map(data_group, function(x){
distance_matrix <- st_distance(x)
biggest_distance <- as.numeric(which(distance_matrix == max(distance_matrix), arr.ind = TRUE)[1,])
farthest_apart <- x[biggest_distance,]
})